don't download sdxl anymore

Usage example - Conditional Rendering Based on App Mode:
```typescript
import { appMode } from "~/signals";

function ConditionalComponent() {
  return (
    <div className="app-container">
      {/* Show/hide components based on current mode */}
      {appMode.value === 'image' && (
        <div className="image-tools">
          <h3>Image Mode Tools</h3>
        </div>
      )}
      
      {appMode.value === 'edit' && (
        <div className="edit-tools">
          <h3>Edit Mode Tools</h3>
        </div>
      )}
  );
}
```

.idea/dictionaries/bt.xml

@@ -2,6 +2,7 @@
   <dictionary name="bt">
     <words>
       <w>arpabet</w>
+      <w>artcraft</w>
       <w>backfill</w>
       <w>backfilling</w>
       <w>backfills</w>

Cargo.toml

@@ -98,6 +98,7 @@ members = [
   # Vendored code
   "crates/vendor/candle/candle-core",
   "crates/vendor/candle/candle-datasets",
+  "crates/vendor/candle/candle-examples",
   "crates/vendor/candle/candle-flash-attn",
   "crates/vendor/candle/candle-kernels",
   "crates/vendor/candle/candle-metal-kernels",

app_config.yaml

@@ -1,7 +1,8 @@
-image_width: 512
-image_height: 512
+image_width: 1024
+image_height: 1024
 #scheduler_steps: 15
 # NB(bt): It looks like we can get down to 2 sampling steps, perhaps ?
-scheduler_steps: 2
+scheduler_steps: 8
 scheduler_samples: 1
-cfg_scale: 1.1
+cfg_scale: 15.0
+seed: 42

crates/desktop/src-tauri/.taurignore

@@ -0,0 +1,4 @@
+# Don't restart the dev server on Rust changes
+crates/
+Cargo.toml
+Cargo.lock

crates/desktop/src-tauri/Cargo.toml

@@ -27,24 +27,56 @@ memory_caching = { path = "../../lib/caching/memory_caching" }
 cudarc = "0.13.7"
 
 # External
+#openssl = { version = "0.10", features = ["vendored"] }
 anyhow = "1.0"
-bytes = "1.10"
 base64 = "0.22.1"
+bytes = "1.10"
 candle-core = { path = "../../vendor/candle/candle-core", features = ["cuda"] }
 candle-nn = { path = "../../vendor/candle/candle-nn", features = ["cuda"] }
 candle-transformers = { path = "../../vendor/candle/candle-transformers", features = ["cuda"] }
+dirs = "6.0.0"
+directories = "6.0.0"
+futures = "0.3"
 hf-hub = "0.3.2"
 image = "0.25.5"
-serde_yaml = "0.9"
 log = "0.4"
 once_cell = "1.20.3"
-rand = "0.8"
+rand = "0.8.5"
+# NB(bt,2025-03-10): default-features = false on Windows while I debug openssl / rustls
+reqwest = { version = "0.12.12", features = ["stream"], default-features = false }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
+serde_yaml = "0.9"
 tauri = { version = "2.2.4", features = [] }
 tauri-plugin-log = "2"
+tempdir = "0.3.7"
+tempfile = "3.18.0"
 tokenizers = "0.15.2"
+tokio = { version = "1.42", features = ["rt", "rt-multi-thread", "fs"] }
+tokio-util = { version = "0.7", features = ["codec"] }
+half = { version = "2.3.1", features = ["num-traits", "use-intrinsics", "rand_distr"] }
 
 [features]
 default = ["cuda"]
-cuda = ["candle-core/cuda", "candle-nn/cuda", "candle-transformers/cuda"]
+cuda = [
+    "candle-core/cuda",
+    "candle-nn/cuda",
+    "candle-transformers/cuda",
+# NB(bt,2025-03-10): Building without Flash attention on Windows for now to bypass linker issues
+    "candle-transformers/flash-attn",
+    "candle-core/cudnn"
+]
+
+[dev-dependencies]
+libloading = "0.8.6"
+serial_test = "3.2.0"
+
+[profile.dev]
+incremental = true # Compile your binary in smaller steps.
+
+[profile.release]
+codegen-units = 1 # Allows LLVM to perform better optimization.
+lto = true # Enables link-time-optimizations.
+opt-level = 3 # `s` prioritizes small binary size. Use `3` if you prefer speed.
+panic = "abort" # Higher performance by disabling panic handlers.
+strip = true # Ensures debug symbols are removed.

crates/desktop/src-tauri/src/endpoints/download_models.rs

@@ -0,0 +1,16 @@
+use crate::ml::downloads::download_all_models::download_all_models;
+use crate::state::app_dir::AppDataRoot;
+use tauri::State;
+
+#[tauri::command]
+pub async fn download_models(
+  app_data_root: State<'_, AppDataRoot>,
+) -> Result<String, String> {
+  
+  download_all_models(&app_data_root).await
+    .map_err(|err| {
+      "error downloading models".to_string()
+    })?;
+  
+  Ok(("downloaded".to_string()))
+}

crates/desktop/src-tauri/src/endpoints/image_endpoint.rs

@@ -1,6 +1,7 @@
+use crate::ml::downloads::download_all_models::download_all_models;
 use crate::ml::model_cache::ModelCache;
 use crate::ml::prompt_cache::PromptCache;
-use crate::ml::stable_diffusion::stable_diffusion_pipeline::{stable_diffusion_pipeline, Args};
+use crate::ml::stable_diffusion::lcm_pipeline::{lcm_pipeline, Args};
 use crate::state::app_config::AppConfig;
 use base64::{engine::general_purpose::STANDARD as BASE64_STANDARD, Engine};
 use bytes::BytesMut;
@@ -9,33 +10,58 @@ use image::{DynamicImage, ImageFormat, ImageReader};
 use log::{error, info};
 use std::io::Cursor;
 use std::path::PathBuf;
-use tauri::State;
+use tauri::{AppHandle, State};
+use crate::state::app_dir::AppDataRoot;
 
 const PROMPT_FILENAME : &str = "prompt.txt";
-const PROMPT: &str = "A beautiful landscape with mountains and a lake";
+const PROMPT: &str = "green goblin, blood dripping from lips, detailed, photorealistic, 8k";
 
 const NEGATIVE_PROMPT: &str = "bad quality, bad faces, poor quality, blurry faces, watermark";
 
+const RANDOM_SEED: u32 = 42;
+
+// TODO: Mark async https://github.com/tauri-apps/tauri/discussions/7737
+
+
 /// This handler takes an image (as a base64 encoded string) and a prompt and returns
 /// an image (as a base64-encoded string).
 #[tauri::command]
-pub fn infer_image(
+pub async fn infer_image(
   image: &str,
   prompt: Option<String>,
   strength: Option<u8>,
-  model_config: State<AppConfig>,
-  model_cache: State<ModelCache>,
-  prompt_cache: State<PromptCache>,
+  model_config: State<'_, AppConfig>,
+  model_cache: State<'_, ModelCache>,
+  prompt_cache: State<'_, PromptCache>,
+  app_data_root: State<'_, AppDataRoot>,
+  app: AppHandle,
 ) -> Result<String, String> {
 
   let prompt = get_prompt_or_fallback(prompt);
 
   info!("Prompt: {}", prompt);
+  info!("Strength: {:?}", strength);
 
   let image = hydrate_base64_image(image)
     .map_err(|err| format!("Couldn't hydrate image from base64: {}", err))?;
 
-  let result = do_infer_image(&prompt, image, strength, &model_config, &model_cache, prompt_cache);
+  // TODO(bt): Move this to another endpoint
+  download_all_models(&app_data_root).await
+    .map_err(|err| {
+      error!("couldn't download models: {:?}", err);
+      "Couldn't download models".to_string()
+    })?;
+
+  let result = do_infer_image(
+    &prompt, 
+    image, 
+    strength, 
+    &model_config, 
+    &model_cache, 
+    prompt_cache, 
+    app, 
+    &app_data_root
+  ).await;
   
   if let Err(err) = result.as_deref() {
     error!("There was an error: {:?}", err);
@@ -44,29 +70,32 @@ pub fn infer_image(
   result
 }
 
-fn do_infer_image(
+async fn do_infer_image(
   prompt: &str,
   image: DynamicImage,
   strength: Option<u8>,
   config: &AppConfig,
   model_cache: &ModelCache,
-  prompt_cache: State<PromptCache>,
+  prompt_cache: State<'_, PromptCache>,
+  app: AppHandle,
+  app_data_root: &AppDataRoot,
 ) -> Result<String, String> {
-  println!("infer_image called; generating image with SDXL Turbo...");
-  
+
   let args = Args {
     image: &image,
     prompt: prompt.to_string(),
     uncond_prompt: NEGATIVE_PROMPT.to_string(),
-    //guidance_scale: Some(0.0),
     model_cache,
     configs: config,
     prompt_cache: &prompt_cache,
     i2i_strength: strength,
     cfg_scale: config.cfg_scale,
+    app: &app,
+    app_data_root,
+    use_flash_attn: true,
   };
 
-  match stable_diffusion_pipeline(args) {
+  match lcm_pipeline(args) {
     Ok(image) => {
       let mut bytes = Vec::with_capacity(1024*1024);
       

crates/desktop/src-tauri/src/endpoints/mod.rs

@@ -1,3 +1,4 @@
+pub mod download_models;
 pub mod flip_image;
 pub mod image_endpoint;
 pub mod test_counter;

crates/desktop/src-tauri/src/events/mod.rs

@@ -0,0 +1 @@
+pub mod notification_event;

crates/desktop/src-tauri/src/events/notification_event.rs

@@ -0,0 +1,30 @@
+use serde::Serialize;
+
+
+#[derive(Clone, Copy, Serialize)]
+#[serde(rename_all = "snake_case")]
+pub enum NotificationModelType {
+  Unet,
+  Vae,
+  Json,
+  ClipEncoder,
+}
+
+#[derive(Clone, Serialize)]
+#[serde(rename_all = "snake_case")]
+pub enum NotificationEvent<'a> {
+  ModelDownloadStarted {
+    model_name: &'a str, 
+    model_type: NotificationModelType,
+  },
+  ModelDownloadProgress {
+    model_name: &'a str,
+    model_type: NotificationModelType,
+    currently_downloaded_bytes: usize,
+    total_file_bytes: usize,
+  },
+  ModelDownloadComplete {
+    model_name: &'a str,
+    model_type: NotificationModelType,
+  }
+}

crates/desktop/src-tauri/src/lib.rs

@@ -1,12 +1,17 @@
 pub mod endpoints;
+pub mod events;
 pub mod ml;
 pub mod state;
+pub mod threads;
+pub mod transfer;
 
+use crate::endpoints::download_models::download_models;
+use crate::endpoints::image_endpoint::infer_image;
+use crate::endpoints::test_counter::test_counter;
 use crate::ml::model_cache::ModelCache;
 use crate::ml::prompt_cache::PromptCache;
 use crate::state::app_config::AppConfig;
-use endpoints::image_endpoint::infer_image;
-use endpoints::test_counter::test_counter;
+use crate::threads::downloader_thread::downloader_thread;
 use tauri_plugin_log::Target;
 use tauri_plugin_log::TargetKind;
 
@@ -24,12 +29,10 @@ pub fn run() {
 
   println!("Creating model cache...");
   
-  let model_cache = ModelCache::new(
-    config.device.clone(),
-    config.dtype,
-    config.sd_version.clone(),
-    config.sd_config.clone()
-  ).expect("Model cache should create");
+  let model_cache = ModelCache::new();
+  
+  let app_data_root = config.app_data_root.clone();
+  let app_data_root2 = config.app_data_root.clone();
 
   println!("Initializing backend runtime...");
 
@@ -46,14 +49,20 @@ pub fn run() {
       //      .build(),
       //  )?;
       //}
+      let app = app.handle().clone();
+
+      tauri::async_runtime::spawn(downloader_thread(app_data_root2, app));
+
       Ok(())
     })
     .manage(config)
     .manage(prompt_cache)
     .manage(model_cache)
+    .manage(app_data_root)
     .invoke_handler(tauri::generate_handler![
       test_counter, 
       infer_image,
+      download_models,
     ])
     .run(tauri::generate_context!())
     .expect("error while running tauri application");

crates/desktop/src-tauri/src/main.rs

@@ -1,6 +1,15 @@
 // Prevents additional console window on Windows in release, DO NOT REMOVE!!
 #![cfg_attr(not(debug_assertions), windows_subsystem = "windows")]
 
+use std::fs::File;
+
 fn main() {
+  //// NB: This is a test to see the behavior of the application with dynamic loading modules.
+  //// We should be able to execute code before loading CUDA and other dependencies.
+  //let r = File::create("program_started.txt");
+  //match r {
+  //  Err(err) => println!("Failed to create program_started.txt: {}", err),
+  //  Ok(_) => {}
+  //}
   app_lib::run();
 }

crates/desktop/src-tauri/src/ml/downloads/download_all_models.rs

@@ -0,0 +1,15 @@
+use crate::ml::downloads::download_model_registry_file::download_model_registry_file;
+use crate::ml::model_type::ModelType;
+use crate::state::app_dir::AppDataRoot;
+
+pub async fn download_all_models(app_data_root: &AppDataRoot) -> anyhow::Result<()> {
+  let weights_dir = app_data_root.weights_dir();
+  
+  //download_model_registry_file(ModelType::ClipJson, weights_dir).await?;
+  //download_model_registry_file(ModelType::SdxlTurboUnet, weights_dir).await?;
+  //download_model_registry_file(ModelType::SdxlTurboVae, weights_dir).await?;
+  //download_model_registry_file(ModelType::SdxlTurboClipEncoder, weights_dir).await?;
+  //download_model_registry_file(ModelType::SdxlTurboClipEncoder2, weights_dir).await?;
+  
+  Ok(())
+}

crates/desktop/src-tauri/src/ml/downloads/download_model_registry_file.rs

@@ -0,0 +1,30 @@
+use crate::ml::model_type::ModelType;
+use crate::state::app_dir::AppWeightsDir;
+use log::info;
+use std::fs::File;
+use std::path::{Path, PathBuf};
+
+pub async fn download_model_registry_file(
+  model_type: ModelType,
+  weights_dir: &AppWeightsDir,
+) -> anyhow::Result<PathBuf> {
+  
+  let path = weights_dir.model_path(&model_type);
+
+  if path.exists() {
+    return Ok(path);
+  }
+
+  info!("downloading model: {:?} to {:?}", model_type, path);
+
+  let body = reqwest::get(model_type.get_download_url())
+    .await?
+    .bytes()
+    .await?;
+  
+  let _file = File::create(&path)?;
+  
+  std::fs::write(&path, body.as_ref())?;
+  
+  Ok(path)
+}

crates/desktop/src-tauri/src/ml/downloads/mod.rs

@@ -0,0 +1,2 @@
+pub mod download_model_registry_file;
+pub mod download_all_models;

crates/desktop/src-tauri/src/ml/mod.rs

@@ -3,4 +3,6 @@ pub mod model_cache;
 pub mod model_file;
 pub mod models;
 pub mod prompt_cache;
-pub mod stable_diffusion;
+pub mod stable_diffusion;
+pub mod model_type;
+pub mod downloads;

crates/desktop/src-tauri/src/ml/model_cache.rs

@@ -1,117 +1,60 @@
 use crate::ml::models::unet_model::UNetModel;
-use crate::ml::model_file::StableDiffusionVersion;
 use anyhow::anyhow;
-use candle_core::{DType, Device, Tensor};
-use candle_transformers::models::stable_diffusion::StableDiffusionConfig;
-use hf_hub::api::sync::Api;
+use candle_transformers::models::stable_diffusion::vae::AutoEncoderKL;
 use std::sync::{Arc, RwLock};
-use candle_transformers::models::stable_diffusion::vae::DiagonalGaussianDistribution;
-use crate::ml::models::lazy_load_vae_model::LazyLoadVaeModel;
+
 // TODO: This data structure is gross. Generalize the locking and loading semantics for reuse.
 
 // Simple registry for now. We can build complex machinery that aids in 
 // VRAM utilization, disk space saving, intelligent scheduling, etc. in the future.
 pub struct ModelCache {
-  // TODO(bt,2025-02-14): Support for more than one device in the future.
-  device: Device,
-  dtype: DType,
-  hf_api: Api,
-  sd_config: StableDiffusionConfig,
-  sd_version: StableDiffusionVersion,
-  
-  unet: Arc<RwLock<Option<UNetModel>>>,
-  lazy_vae: LazyLoadVaeModel,
+  vae: RwLock<Option<Arc<AutoEncoderKL>>>,
+  unet2: RwLock<Option<Arc<UNetModel>>>,
 }
 
 impl ModelCache {
-  pub fn new(
-    device: Device, 
-    dtype: DType,
-    sd_version: StableDiffusionVersion, 
-    sd_config: StableDiffusionConfig,
-  ) -> anyhow::Result<Self> {
-    let api = Api::new()?;
-    
-    Ok(Self { 
-      device: device.clone(),
-      dtype,
-      hf_api: api.clone(),
-      unet: Arc::new(RwLock::new(None)),
-      lazy_vae: LazyLoadVaeModel::new(
-        &sd_config,
-        &sd_version,
-        &api,
-        &device,
-        dtype,
-      )?,
-      sd_config,
-      sd_version,
-    })
-  }
-
-  pub fn vae_encode(&self, xs: &Tensor) -> anyhow::Result<DiagonalGaussianDistribution> {
-    self.lazy_vae.encode(xs)
-  }
-
-  pub fn vae_decode(&self, xs: &Tensor) -> anyhow::Result<Tensor> {
-    self.lazy_vae.decode(xs)
-  }
-  
-  pub fn unet_inference(&self, latent_model_input: &Tensor, timestep: f64, text_embeddings: &Tensor) -> anyhow::Result<Tensor> {
-    println!("unet_inference");
-    match self.try_unet_inference(latent_model_input, timestep, text_embeddings) {
-      Err(err) => Err(err),
-      Ok(Some(result)) => Ok(result),
-      Ok(None) => {
-        // Model wasn't previously loaded
-        self.load_unet()?;
-        self.try_unet_inference(latent_model_input, timestep, text_embeddings)?
-          .ok_or(anyhow!("model was not previously loaded"))
-      }
+  pub fn new() -> Self {
+    Self { 
+      vae: RwLock::new(None),
+      unet2: RwLock::new(None),
     }
   }
   
-  fn load_unet(&self) -> anyhow::Result<()> {
-    println!("load_unet");
-    match self.unet.write() {
-      Err(err) => return Err(anyhow!("lock error: {:?}", err)),
-      Ok(mut maybe_model) => {
-        match &*maybe_model {
-          Some(_model) => {} // Fall through
-          None => {
-            let repo = self.sd_version.repo();
-            
-            println!("Downloading UNET model files from: {} ...", repo);
-
-            let unet_file = self.hf_api.model(repo.to_string())
-              .get("unet/diffusion_pytorch_model.safetensors")
-              .map_err(|err| anyhow!("error fetching model: {:?}", err))?;
-
-            println!("Building UNET model... (3)");
-            let unet = UNetModel::new(&self.sd_config, unet_file, &self.device, self.dtype)
-              .map_err(|err| anyhow!("error initializing unet model: {:?}", err))?;
-
-            println!("Built UNET...");
-            *maybe_model = Some(unet);
-          }
-        }
+  pub fn set_vae(&self, vae: Arc<AutoEncoderKL>) -> anyhow::Result<()> {
+    match self.vae.write() {
+      Err(err) => Err(anyhow!("{}", err)),
+      Ok(mut lock) => {
+        *lock = Some(vae);
+        Ok(())
       }
     }
-    Ok(())
   }
 
-  pub fn try_unet_inference(&self, latent_model_input: &Tensor, timestep: f64, text_embeddings: &Tensor) -> anyhow::Result<Option<Tensor>> {
-    println!("try_unet_inference");
-    match self.unet.read() {
-      Ok(mut model) => {
-        match &*model {
-          None => Ok(None), // Model is not yet loaded
-          Some(model) => {
-            Ok(Some(model.inference(latent_model_input, timestep, text_embeddings)?))
-          }
-        }
+  pub fn get_vae(&self) -> anyhow::Result<Option<Arc<AutoEncoderKL>>> {
+    match self.vae.read() {
+      Err(err) => Err(anyhow!("{}", err)),
+      Ok(ref lock) => {
+        Ok(lock.as_ref().map(|v|v.clone()))
+      }
+    }
+  }
+
+  pub fn set_unet(&self, vae: Arc<UNetModel>) -> anyhow::Result<()> {
+    match self.unet2.write() {
+      Err(err) => Err(anyhow!("{}", err)),
+      Ok(mut lock) => {
+        *lock = Some(vae);
+        Ok(())
+      }
+    }
+  }
+
+  pub fn get_unet(&self) -> anyhow::Result<Option<Arc<UNetModel>>> {
+    match self.unet2.read() {
+      Err(err) => Err(anyhow!("{}", err)),
+      Ok(ref lock) => {
+        Ok(lock.as_ref().map(|v|v.clone()))
       }
-      Err(err) => Err(anyhow!("lock error: {:?}", err)),
     }
   }
   

crates/desktop/src-tauri/src/ml/model_file.rs

@@ -4,80 +4,177 @@ extern crate accelerate_src;
 extern crate intel_mkl_src;
 
 use candle_transformers::models::stable_diffusion;
+use anyhow::{Error as E, Result};
+use hf_hub::api::sync::Api;
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum StableDiffusionVersion {
+    V1_5,
+    V1_5Inpaint,
+    V2_1,
+    V2Inpaint,
+    Xl,
+    XlInpaint,
+    Turbo,
+}
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum ModelFile {
-  Tokenizer,
-  Tokenizer2,
-  Clip,
-  Clip2,
-  Unet,
-  Vae,
-}
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
-pub enum StableDiffusionVersion {
-  V1_5,
-  V1_5Inpaint,
-  V2_1,
-  V2Inpaint,
-  Xl,
-  XlInpaint,
-  Turbo,
+    Tokenizer,
+    Tokenizer2,
+    Clip,
+    Clip2,
+    Unet,
+    UnetLcm,
+    Vae,
 }
 
 impl StableDiffusionVersion {
-  pub fn repo(&self) -> &'static str {
-    match self {
-      Self::XlInpaint => "diffusers/stable-diffusion-xl-1.0-inpainting-0.1",
-      Self::Xl => "stabilityai/stable-diffusion-xl-base-1.0",
-      Self::V2Inpaint => "stabilityai/stable-diffusion-2-inpainting",
-      Self::V2_1 => "stabilityai/stable-diffusion-2-1",
-      Self::V1_5 => "runwayml/stable-diffusion-v1-5",
-      Self::V1_5Inpaint => "stable-diffusion-v1-5/stable-diffusion-inpainting",
-      Self::Turbo => "stabilityai/sdxl-turbo",
+    pub fn repo(&self) -> &'static str {
+        match self {
+            Self::XlInpaint => "diffusers/stable-diffusion-xl-1.0-inpainting-0.1",
+            Self::Xl => "stabilityai/stable-diffusion-xl-base-1.0",
+            Self::V2Inpaint => "stabilityai/stable-diffusion-2-inpainting",
+            Self::V2_1 => "stabilityai/stable-diffusion-2-1",
+            Self::V1_5 => "Lykon/dreamshaper-7",
+            Self::V1_5Inpaint => "stable-diffusion-v1-5/stable-diffusion-inpainting",
+            Self::Turbo => "Lykon/dreamshaper-7",
+        }
+    }
+
+    fn unet_file(&self, use_f16: bool) -> &'static str {
+        match self {
+            Self::V1_5
+            | Self::V1_5Inpaint
+            | Self::V2_1
+            | Self::V2Inpaint
+            | Self::Xl
+            | Self::XlInpaint
+            | Self::Turbo => {
+                if use_f16 {
+                    "unet/diffusion_pytorch_model.fp16.safetensors"
+                } else {
+                    "unet/diffusion_pytorch_model.safetensors"
+                }
+            }
+        }
+    }
+
+    fn vae_file(&self, use_f16: bool) -> &'static str {
+        match self {
+            Self::V1_5
+            | Self::V1_5Inpaint
+            | Self::V2_1
+            | Self::V2Inpaint
+            | Self::Xl
+            | Self::XlInpaint
+            | Self::Turbo => {
+                if use_f16 {
+                    "vae/diffusion_pytorch_model.fp16.safetensors"
+                } else {
+                    "vae/diffusion_pytorch_model.safetensors"
+                }
+            }
+        }
+    }
+
+    fn clip_file(&self, use_f16: bool) -> &'static str {
+        match self {
+            Self::V1_5
+            | Self::V1_5Inpaint
+            | Self::V2_1
+            | Self::V2Inpaint
+            | Self::Xl
+            | Self::XlInpaint
+            | Self::Turbo => {
+                if use_f16 {
+                    "text_encoder/model.fp16.safetensors"
+                } else {
+                    "text_encoder/model.safetensors"
+                }
+            }
+        }
+    }
+
+    fn clip2_file(&self, use_f16: bool) -> &'static str {
+        match self {
+            Self::V1_5
+            | Self::V1_5Inpaint
+            | Self::V2_1
+            | Self::V2Inpaint
+            | Self::Xl
+            | Self::XlInpaint
+            | Self::Turbo => {
+                if use_f16 {
+                    "text_encoder_2/model.fp16.safetensors"
+                } else {
+                    "text_encoder_2/model.safetensors"
+                }
+            }
+        }
     }
-  }
 }
 
 impl ModelFile {
-  pub fn get(
-    &self,
-    filename: Option<String>,
-    version: StableDiffusionVersion,
-    use_f16: bool,
-  ) -> anyhow::Result<std::path::PathBuf> {
-    use hf_hub::api::sync::Api;
-    match filename {
-      Some(filename) => Ok(std::path::PathBuf::from(filename)),
-      None => {
-        let (repo, path) = match self {
-          Self::Tokenizer => {
-            let tokenizer_repo = match version {
-              StableDiffusionVersion::V1_5
-              | StableDiffusionVersion::V2_1
-              | StableDiffusionVersion::V1_5Inpaint
-              | StableDiffusionVersion::V2Inpaint => "openai/clip-vit-base-patch32",
-              StableDiffusionVersion::Xl
-              | StableDiffusionVersion::XlInpaint
-              | StableDiffusionVersion::Turbo => {
-                "openai/clip-vit-large-patch14"
-              }
-            };
-            (tokenizer_repo, "tokenizer.json")
-          }
-          Self::Tokenizer2 => {
-            ("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k", "tokenizer.json")
-          }
-          Self::Clip => (version.repo(), "text_encoder/model.safetensors"),
-          Self::Clip2 => (version.repo(), "text_encoder_2/model.safetensors"),
-          Self::Unet => (version.repo(), "unet/diffusion_pytorch_model.safetensors"),
-          Self::Vae => (version.repo(), "vae/diffusion_pytorch_model.safetensors"),
-        };
-        let filename = Api::new()?.model(repo.to_string()).get(path)?;
-        Ok(filename)
-      }
+    pub fn get(
+        &self,
+        filename: Option<String>,
+        version: StableDiffusionVersion,
+        use_f16: bool,
+    ) -> Result<std::path::PathBuf> {
+        match filename {
+            Some(filename) => Ok(std::path::PathBuf::from(filename)),
+            None => {
+                let (repo, path) = match self {
+                    Self::Tokenizer => {
+                        let tokenizer_repo = match version {
+                            StableDiffusionVersion::V1_5
+                            | StableDiffusionVersion::V2_1
+                            | StableDiffusionVersion::V1_5Inpaint
+                            | StableDiffusionVersion::V2Inpaint => "openai/clip-vit-base-patch32",
+                            StableDiffusionVersion::Xl
+                            | StableDiffusionVersion::XlInpaint
+                            | StableDiffusionVersion::Turbo => {
+                                // This seems similar to the patch32 version except some very small
+                                // difference in the split regex.
+                                "openai/clip-vit-large-patch14"
+                            }
+                        };
+                        (tokenizer_repo, "tokenizer.json")
+                    }
+                    Self::Tokenizer2 => {
+                        ("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k", "tokenizer.json")
+                    }
+                    Self::Clip => (version.repo(), version.clip_file(use_f16)),
+                    Self::Clip2 => (version.repo(), version.clip2_file(use_f16)),
+                    Self::UnetLcm => (
+                        "SimianLuo/LCM_Dreamshaper_v7",
+                        if use_f16 {
+                            "unet/diffusion_pytorch_model.fp16.safetensors"
+                        } else {
+                            "unet/diffusion_pytorch_model.safetensors"
+                        },
+                    ),
+                    Self::Unet => (version.repo(), version.unet_file(use_f16)),
+                    Self::Vae => {
+                        // Override for SDXL when using f16 weights.
+                        // See https://github.com/huggingface/candle/issues/1060
+                        if matches!(
+                            version,
+                            StableDiffusionVersion::Xl | StableDiffusionVersion::Turbo,
+                        ) && use_f16
+                        {
+                            (
+                                "madebyollin/sdxl-vae-fp16-fix",
+                                "diffusion_pytorch_model.safetensors",
+                            )
+                        } else {
+                            (version.repo(), version.vae_file(use_f16))
+                        }
+                    }
+                };
+                let filename = Api::new()?.model(repo.to_string()).get(path)?;
+                Ok(filename)
+            }
+        }
     }
-  }
 }
-

crates/desktop/src-tauri/src/ml/model_type.rs

@@ -0,0 +1,86 @@
+use crate::events::notification_event::NotificationModelType;
+use crate::state::app_dir::AppWeightsDir;
+use std::path::PathBuf;
+
+#[derive(Copy, Clone, Debug)]
+pub enum ModelType {
+  ClipJson,
+  SdxlTurboUnet,
+  SdxlTurboVae,
+  SdxlTurboClipEncoder,
+  SdxlTurboClipEncoder2,
+  SimianLuoLcmDreamshaperV7Unet,
+  LykonDreamshaper7Vae,
+  LykonDreamshaper7TextEncoderFp16,
+}
+
+impl ModelType {
+  pub fn get_name(&self) -> &'static str {
+    match self {
+      Self::ClipJson => "CLIP (JSON)",
+      Self::SdxlTurboUnet => "SDXL Turbo UNET",
+      Self::SdxlTurboVae => "SDXL Turbo VAE",
+      Self::SdxlTurboClipEncoder => "Clip Encoder #1",
+      Self::SdxlTurboClipEncoder2 => "Clip Encoder #2",
+      Self::SimianLuoLcmDreamshaperV7Unet => "SimianLuo LCM Dreamshaper v7 UNET",
+      Self::LykonDreamshaper7Vae => "Lykon Dreamshaper 7 VAE",
+      Self::LykonDreamshaper7TextEncoderFp16 => "Lykon Dreamshaper Text Encoder (fp16)",
+    }
+  }
+
+  pub fn get_notification_type(&self) -> NotificationModelType {
+    match self {
+      Self::ClipJson => NotificationModelType::Json,
+      Self::SdxlTurboUnet => NotificationModelType::Unet,
+      Self::SdxlTurboVae => NotificationModelType::Vae,
+      Self::SdxlTurboClipEncoder => NotificationModelType::ClipEncoder,
+      Self::SdxlTurboClipEncoder2 => NotificationModelType::ClipEncoder,
+      Self::SimianLuoLcmDreamshaperV7Unet => NotificationModelType::Unet,
+      Self::LykonDreamshaper7Vae => NotificationModelType::Vae,
+      Self::LykonDreamshaper7TextEncoderFp16 => NotificationModelType::ClipEncoder,
+    }
+  }
+
+  pub fn get_hf_id(&self) -> Option<&'static str> {
+    match self {
+      Self::ClipJson => Some("laion/CLIP-ViT-bigG-14-laion2B-39B-b160k"),
+      Self::SdxlTurboUnet 
+      | Self::SdxlTurboVae
+      | Self::SdxlTurboClipEncoder
+      | Self::SdxlTurboClipEncoder2 => Some("stabilityai/sdxl-turbo"),
+      Self::SimianLuoLcmDreamshaperV7Unet => Some("SimianLuo/LCM_Dreamshaper_v7"),
+      Self::LykonDreamshaper7Vae
+      | Self::LykonDreamshaper7TextEncoderFp16 => Some("Lykon/dreamshaper-7"),
+    }
+  }
+
+  pub fn get_download_url(&self) -> &'static str {
+    match self {
+      Self::ClipJson => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/tokenizer.json",
+      Self::SdxlTurboUnet => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/diffusion_pytorch_model.unet.safetensors",
+      Self::SdxlTurboVae => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/diffusion_pytorch_model.vae.safetensors",
+      Self::SdxlTurboClipEncoder => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/clip_text_encoder.safetensors",
+      Self::SdxlTurboClipEncoder2 => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/clip_text_encoder_2.safetensors",
+      Self::SimianLuoLcmDreamshaperV7Unet => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/simianluo_lcm_dreamshaper_v7_unet.safetensors",
+      Self::LykonDreamshaper7Vae => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/lykon_dreamshaper_7_vae.safetensors",
+      Self::LykonDreamshaper7TextEncoderFp16 => "https://pub-bc5e2bc0cdee4bb5ae8fca9d641ca0d6.r2.dev/lykon_dreamshaper_7_text_encoder.fp16.safetensors",
+    }
+  }
+
+  pub fn get_filename(&self) -> &'static str {
+    match self {
+      Self::ClipJson => "clip_vit_bigg_14_lion2b_39b_b160k.tokenizer.json",
+      Self::SdxlTurboUnet => "diffusion_pytorch_model.unet.safetensors",
+      Self::SdxlTurboVae => "diffusion_pytorch_model.vae.safetensors",
+      Self::SdxlTurboClipEncoder => "clip_text_encoder.safetensors",
+      Self::SdxlTurboClipEncoder2 => "clip_text_encoder_2.safetensors",
+      Self::SimianLuoLcmDreamshaperV7Unet => "simianluo_lcm_dreamshaper_v7_unet.safetensors",
+      Self::LykonDreamshaper7Vae => "lykon_dreamshaper_7_vae.safetensors",
+      Self::LykonDreamshaper7TextEncoderFp16 => "lykon_dreamshaper_7_text_encoder.fp16.safetensors",
+    }
+  }
+  
+  pub fn get_path(&self, weights_dir: &AppWeightsDir) -> PathBuf {
+    weights_dir.model_path(self)
+  }
+}

crates/desktop/src-tauri/src/ml/models/unet_model.rs

@@ -8,28 +8,23 @@ pub struct UNetModel {
 }
 
 impl UNetModel {
-  pub fn new<P: AsRef<Path>>(
-    sd_config: &StableDiffusionConfig,
-    unet_weights_file: P,
-    device: &Device,
-    dtype: DType,
-  ) -> anyhow::Result<Self> {
+  pub fn new<P: AsRef<Path>>(sd_config: &StableDiffusionConfig, unet_weights_file: P, device: &Device, dtype: DType) -> anyhow::Result<Self> {
     println!("building unet model... (1)");
-    
-    let model = sd_config.build_unet(
-      unet_weights_file, 
-      &device, 
-      4, 
-      false, 
-      dtype
-    )?;
-    
-    Ok(Self {
-      model,
-    })
+
+    let model = sd_config.build_unet(unet_weights_file, &device, 4, true, dtype)?;
+
+    Ok(Self { model })
   }
-  
+
+  pub fn new_with_inner(model: UNet2DConditionModel) -> Self {
+    Self { model }
+  }
+
   pub fn inference(&self, latent_model_input: &Tensor, timestep: f64, text_embeddings: &Tensor) -> anyhow::Result<Tensor> {
     Ok(self.model.forward(&latent_model_input, timestep, &text_embeddings)?)
   }
+
+  pub fn forward_with_guidance(&self, latent_model_input: &Tensor, timestep: f64, text_embeddings: &Tensor, guidance_scale_embedding: &Tensor) -> anyhow::Result<Tensor> {
+    Ok(self.model.forward_with_guidance(&latent_model_input, timestep, &text_embeddings, Some(guidance_scale_embedding))?)
+  }
 }

crates/desktop/src-tauri/src/ml/stable_diffusion/infer_clip_text_embeddings.rs

@@ -3,6 +3,7 @@ extern crate accelerate_src;
 #[cfg(feature = "mkl")]
 extern crate intel_mkl_src;
 
+use std::path::PathBuf;
 use candle_transformers::models::stable_diffusion;
 
 use crate::ml::model_file::{ModelFile, StableDiffusionVersion};
@@ -10,27 +11,15 @@ use anyhow::Error as E;
 use candle_core::{DType, Device, Module, Tensor, D};
 use log::info;
 use tokenizers::Tokenizer;
+use crate::ml::model_type::ModelType;
+use crate::state::app_dir::AppWeightsDir;
 
 #[allow(clippy::too_many_arguments)]
-pub fn infer_clip_text_embeddings(
-  prompt: &str,
-  uncond_prompt: &str,
-  tokenizer: Option<String>,
-  clip_weights: Option<String>,
-  clip2_weights: Option<String>,
-  sd_version: StableDiffusionVersion,
-  sd_config: &stable_diffusion::StableDiffusionConfig,
-  use_f16: bool,
-  device: &Device,
-  dtype: DType,
-  use_guide_scale: bool,
-) -> anyhow::Result<Tensor> {
-  info!("Preparing text embeddings...");
+pub fn infer_clip_text_embeddings(prompt: &str, uncond_prompt: &str, tokenizer: Option<String>, clip_weights: Option<String>, clip2_weights: Option<String>, sd_version: StableDiffusionVersion, sd_config: &stable_diffusion::StableDiffusionConfig, use_f16: bool, device: &Device, dtype: DType, use_guide_scale: bool, weights_dir: &AppWeightsDir) -> anyhow::Result<Tensor> {
+  info!("Preparing text embeddings... for {:?}", sd_version);
 
   let which = match sd_version {
-    StableDiffusionVersion::Xl
-    | StableDiffusionVersion::XlInpaint
-    | StableDiffusionVersion::Turbo => vec![true, false],
+    StableDiffusionVersion::Xl | StableDiffusionVersion::XlInpaint | StableDiffusionVersion::Turbo => vec![true, false],
     _ => vec![true],
   };
 
@@ -41,7 +30,7 @@ pub fn infer_clip_text_embeddings(
         &prompt,
         &uncond_prompt,
         None, // tokenizer
-        None, // clip_weights
+        clip_weights.clone(),
         None, // clip2_weights
         sd_version,
         &sd_config,
@@ -50,55 +39,45 @@ pub fn infer_clip_text_embeddings(
         dtype,
         use_guide_scale,
         *first,
+        weights_dir,
       )
     })
     .collect::<anyhow::Result<Vec<_>>>()?;
 
   let text_embeddings = Tensor::cat(&text_embeddings, D::Minus1)?;
-  
+
   Ok(text_embeddings)
 }
 
 #[allow(clippy::too_many_arguments)]
-fn do_infer_clip_text_embeddings(
-    prompt: &str,
-    uncond_prompt: &str,
-    tokenizer: Option<String>,
-    clip_weights: Option<String>,
-    clip2_weights: Option<String>,
-    sd_version: StableDiffusionVersion,
-    sd_config: &stable_diffusion::StableDiffusionConfig,
-    use_f16: bool,
-    device: &Device,
-    dtype: DType,
-    use_guide_scale: bool,
-    first: bool,
-  ) -> anyhow::Result<Tensor> {
-  
-  let tokenizer_file = if first {
-    ModelFile::Tokenizer
-  } else {
-    ModelFile::Tokenizer2
-  };
-  let tokenizer = tokenizer_file.get(tokenizer, sd_version, use_f16)?;
+fn do_infer_clip_text_embeddings(prompt: &str, uncond_prompt: &str, tokenizer: Option<String>, clip_weights: Option<String>, clip2_weights: Option<String>, sd_version: StableDiffusionVersion, sd_config: &stable_diffusion::StableDiffusionConfig, use_f16: bool, device: &Device, dtype: DType, use_guide_scale: bool, first: bool, weights_dir: &AppWeightsDir) -> anyhow::Result<Tensor> {
+  info!("do_infer_clip_text_embeddings called with args {:?}", (prompt, uncond_prompt, tokenizer.clone().unwrap_or_else(|| "None".to_string()), clip_weights.clone().unwrap_or_else(|| "None".to_string()), clip2_weights.clone().unwrap_or_else(|| "None".to_string()), sd_version, sd_config, use_f16, device, dtype, use_guide_scale, first));
+
+  //let tokenizer_file = if first {
+  //  info!("ModelFile::Tokenizer");
+  //  ModelFile::Tokenizer
+  //} else {
+  //  info!("ModelFile::Tokenizer2");
+  //  ModelFile::Tokenizer2
+  //};
+
+  //let tokenizer = tokenizer_file.get(tokenizer, sd_version, use_f16)?;
+
+  let tokenizer = weights_dir.model_path(&ModelType::ClipJson);
+
+  info!("Loading Clip Tokenizer path: {:?}", tokenizer);
+
   let tokenizer = Tokenizer::from_file(tokenizer).map_err(E::msg)?;
+
   let pad_id = match &sd_config.clip.pad_with {
     Some(padding) => *tokenizer.get_vocab(true).get(padding.as_str()).unwrap(),
     None => *tokenizer.get_vocab(true).get("<|endoftext|>").unwrap(),
   };
 
   println!("Running with prompt \"{prompt}\".");
-  let mut tokens = tokenizer
-    .encode(prompt, true)
-    .map_err(E::msg)?
-    .get_ids()
-    .to_vec();
+  let mut tokens = tokenizer.encode(prompt, true).map_err(E::msg)?.get_ids().to_vec();
   if tokens.len() > sd_config.clip.max_position_embeddings {
-    anyhow::bail!(
-            "the prompt is too long, {} > max-tokens ({})",
-            tokens.len(),
-            sd_config.clip.max_position_embeddings
-        )
+    anyhow::bail!("the prompt is too long, {} > max-tokens ({})", tokens.len(), sd_config.clip.max_position_embeddings)
   }
   while tokens.len() < sd_config.clip.max_position_embeddings {
     tokens.push(pad_id)
@@ -106,37 +85,23 @@ fn do_infer_clip_text_embeddings(
   let tokens = Tensor::new(tokens.as_slice(), device)?.unsqueeze(0)?;
 
   println!("Building the Clip transformer.");
-  let clip_weights_file = if first {
-    ModelFile::Clip
-  } else {
-    ModelFile::Clip2
-  };
+  let clip_weights_file = if first { ModelFile::Clip } else { ModelFile::Clip2 };
   let clip_weights = if first {
-    clip_weights_file.get(clip_weights, sd_version, use_f16)?
+    //clip_weights_file.get(clip_weights, sd_version, use_f16)?
+    weights_dir.model_path(&ModelType::SdxlTurboClipEncoder)
   } else {
-    clip_weights_file.get(clip2_weights, sd_version, use_f16)?
-  };
-  let clip_config = if first {
-    &sd_config.clip
-  } else {
-    sd_config.clip2.as_ref().unwrap()
+    //clip_weights_file.get(clip2_weights, sd_version, use_f16)?
+    weights_dir.model_path(&ModelType::SdxlTurboClipEncoder2)
   };
+  let clip_config = if first { &sd_config.clip } else { sd_config.clip2.as_ref().unwrap() };
   let text_model = stable_diffusion::build_clip_transformer(clip_config, clip_weights, device, DType::F32)?;
 
   let text_embeddings = text_model.forward(&tokens)?;
 
   let text_embeddings = if use_guide_scale {
-    let mut uncond_tokens = tokenizer
-      .encode(uncond_prompt, true)
-      .map_err(E::msg)?
-      .get_ids()
-      .to_vec();
+    let mut uncond_tokens = tokenizer.encode(uncond_prompt, true).map_err(E::msg)?.get_ids().to_vec();
     if uncond_tokens.len() > sd_config.clip.max_position_embeddings {
-      anyhow::bail!(
-                "the negative prompt is too long, {} > max-tokens ({})",
-                uncond_tokens.len(),
-                sd_config.clip.max_position_embeddings
-            )
+      anyhow::bail!("the negative prompt is too long, {} > max-tokens ({})", uncond_tokens.len(), sd_config.clip.max_position_embeddings)
     }
     while uncond_tokens.len() < sd_config.clip.max_position_embeddings {
       uncond_tokens.push(pad_id)

crates/desktop/src-tauri/src/ml/stable_diffusion/lcm_pipeline.rs

@@ -0,0 +1,291 @@
+use std::path::PathBuf;
+use std::sync::{Arc, RwLock, RwLockReadGuard, RwLockWriteGuard};
+
+use crate::events::notification_event::{NotificationEvent, NotificationModelType};
+use crate::ml::image::dynamic_image_to_tensor::dynamic_image_to_tensor;
+use crate::ml::image::tensor_to_image_buffer::{tensor_to_image_buffer, RgbImage};
+use crate::ml::model_cache::ModelCache;
+use crate::ml::model_file::{ModelFile, StableDiffusionVersion};
+use crate::ml::model_type::ModelType;
+use crate::ml::models::unet_model::UNetModel;
+use crate::ml::prompt_cache::PromptCache;
+use crate::ml::stable_diffusion::get_vae_scale::get_vae_scale;
+use crate::ml::stable_diffusion::infer_clip_text_embeddings::infer_clip_text_embeddings;
+use crate::state::app_config::AppConfig;
+use crate::state::app_dir::AppDataRoot;
+use anyhow::{anyhow, Error as E, Result};
+use candle_core::{DType, IndexOp, Tensor, D};
+use candle_nn::VarBuilder;
+use candle_transformers::models::stable_diffusion::lcm::LCMScheduler;
+use candle_transformers::models::stable_diffusion::unet_2d::BlockConfig;
+use candle_transformers::models::stable_diffusion::unet_2d::UNet2DConditionModel;
+use candle_transformers::models::stable_diffusion::unet_2d::UNet2DConditionModelConfig;
+use candle_transformers::models::stable_diffusion::vae::{AutoEncoderKL, DiagonalGaussianDistribution};
+use image::DynamicImage;
+use log::info;
+use rand::Rng;
+use tauri::{AppHandle, Emitter};
+
+pub struct Args<'a> {
+  pub image: &'a DynamicImage,
+  pub prompt: String,
+  pub uncond_prompt: String,
+  pub cfg_scale: Option<f64>,
+  pub i2i_strength: Option<u8>,
+  pub configs: &'a AppConfig,
+  pub model_cache: &'a ModelCache,
+  pub prompt_cache: &'a PromptCache,
+  pub app_data_root: &'a AppDataRoot,
+  pub app: &'a AppHandle,
+  pub use_flash_attn: bool,
+}
+
+pub fn lcm_pipeline(args: Args<'_>) -> Result<RgbImage> {
+  let Args { prompt, uncond_prompt, cfg_scale, i2i_strength, configs, model_cache, prompt_cache, app, image, app_data_root, use_flash_attn } = args;
+
+  let weights_dir = app_data_root.weights_dir();
+
+  let img2img_strength = args.i2i_strength.unwrap_or(75) as f64 / 100.0;
+  let use_f16 = true;
+
+  //println!("Starting image generation with the following configuration:");
+  //println!("  Model: {:?}", configs.sd_version);
+  //println!("  Prompt: {}", prompt);
+  //println!("  Steps: {}", configs.scheduler_steps);
+  //println!("  Device: {:?}", configs.device);
+
+  //println!("Model dimensions: {}x{}", configs.sd_config.width, configs.sd_config.height);
+
+  // Use LCM Scheduler instead of Euler Ancestral for better speed and quality
+  let mut scheduler = LCMScheduler::new(configs.scheduler_steps, img2img_strength, candle_transformers::models::stable_diffusion::lcm::LCMSchedulerConfig::default())?;
+
+  let seed = configs.seed.unwrap_or_else(|| rand::thread_rng().gen());
+  configs.device.set_seed(seed)?;
+
+  //info!("Using seed: {}", seed);
+
+  let guidance_scale = match cfg_scale {
+    Some(guidance_scale) => guidance_scale,
+    None => match configs.sd_version {
+      StableDiffusionVersion::V1_5 | StableDiffusionVersion::V2_1 | StableDiffusionVersion::Xl => 7.5,
+      StableDiffusionVersion::Turbo => 0.,
+      _ => 0., // NB: Not sure what the other model families should use, so sticking with "0"
+    },
+  };
+
+  // let use_guide_scale = guidance_scale > 1.0;
+  let use_guide_scale = false;
+
+  //info!("Using guide scale: {}", use_guide_scale);
+
+  //info!("Checking if prompt is cached");
+  let maybe_cached = prompt_cache.get_copy(&prompt)?;
+
+  let mut text_embeddings = if let Some(tensor) = maybe_cached {
+    tensor
+  } else {
+    info!("Loading clip weights...");
+    //println!("Loading clip weights for {:?} with model {:?}", configs.sd_version, configs.sd_version.repo());
+    //let api = configs.hf_api.clone();
+
+    //let clip_path = api.model(configs.sd_version.repo().to_string()).get(if use_f16 { "text_encoder/model.fp16.safetensors" } else { "text_encoder/model.safetensors" })?;
+
+    //println!(">>>> CLIP REPO = {:?}", configs.sd_version.repo());
+    //println!(">>>> CLIP PATH = {:?}", clip_path);
+    //println!(">>>> USE_FP16 = {:?}", use_f16);
+    //let clip_weights = Some(clip_path.to_string_lossy().to_string());
+
+    let clip_weights = weights_dir.model_path(&ModelType::LykonDreamshaper7TextEncoderFp16);
+    let clip_weights = Some(clip_weights.to_string_lossy().to_string());
+
+    info!("Prompt is NOT cached! Calculating embedding...");
+    let tensor = infer_clip_text_embeddings(
+      &prompt,
+      &uncond_prompt,
+      None, // tokenizer
+      clip_weights,
+      None, // clip2_weights
+      configs.sd_version,
+      &configs.sd_config,
+      use_f16, // use_f16
+      &configs.device,
+      configs.dtype,
+      use_guide_scale,
+      weights_dir,
+    )?;
+    prompt_cache.store_copy(&prompt, &tensor)?;
+    tensor
+  };
+
+  println!("Text embeddings shape: {:?}", text_embeddings.shape());
+
+  println!("Loading input image into tensor...");
+
+  let input_image = dynamic_image_to_tensor(image, &configs.device, configs.dtype)?;
+
+  println!("Reference image shape: {:?}", input_image.shape());
+
+  let vae_scale = get_vae_scale(configs.sd_version);
+
+  let maybe_vae = model_cache.get_vae()?;
+
+  let vae = match maybe_vae {
+    Some(vae) => vae,
+    None => {
+      //let repo = configs.sd_version.repo();
+      //println!(">>>>> VAE REPO = {:?}", repo);
+      //println!("Building VAE model from : {:?} ... (3)", repo);
+      //let vae_file = configs.hf_api.model(repo.to_string()).get("vae/diffusion_pytorch_model.safetensors")?;
+
+      let vae_file = weights_dir.model_path(&ModelType::LykonDreamshaper7Vae);
+
+      println!("Building VAE model from file {:?}...", &vae_file);
+
+      let vae = configs.sd_config.build_vae(vae_file, &configs.device, configs.dtype)?;
+
+      let vae = Arc::new(vae);
+
+      model_cache.set_vae(vae.clone())?;
+
+      vae
+    }
+  };
+
+  let maybe_unet = model_cache.get_unet()?;
+
+  let unet = match maybe_unet {
+    Some(unet) => unet,
+    None => {
+      info!("No unet found in cache; loading...");
+
+      let unet_weights = weights_dir.model_path(&ModelType::SimianLuoLcmDreamshaperV7Unet);
+
+      let in_channels = match configs.sd_version {
+        StableDiffusionVersion::XlInpaint | StableDiffusionVersion::V2Inpaint | StableDiffusionVersion::V1_5Inpaint => 9,
+        _ => 4,
+      };
+
+      // For LCM models, ensure that:
+      // 1. We're using the correct prediction type (epsilon)
+      // 2. The cross-attention dimension is correct (768 for Dreamshaper)
+      // 3. The in_channels matches the expected input (4 for standard, 9 for inpainting)
+      let unet_config = UNet2DConditionModelConfig {
+        // LCM specific config values
+        cross_attention_dim: 768,      // Dreamshaper v7 uses 768
+        time_cond_proj_dim: Some(256), // LCM models use 256-dimensional guidance embeddings
+        // Rest of configuration...
+        center_input_sample: false,
+        flip_sin_to_cos: true,
+        freq_shift: 0.0,
+        blocks: vec![
+          BlockConfig { out_channels: 320, use_cross_attn: Some(1), attention_head_dim: 8 },
+          BlockConfig { out_channels: 640, use_cross_attn: Some(1), attention_head_dim: 8 },
+          BlockConfig { out_channels: 1280, use_cross_attn: Some(1), attention_head_dim: 8 },
+          BlockConfig { out_channels: 1280, use_cross_attn: None, attention_head_dim: 8 }
+        ],
+        layers_per_block: 2,
+        downsample_padding: 1,
+        mid_block_scale_factor: 1.0,
+        norm_num_groups: 32,
+        norm_eps: 1e-5,
+        sliced_attention_size: None,
+        use_linear_projection: false,
+        time_embed_dim: Some(1280),
+        // time_embed_dim: Some(256),
+      };
+
+      // Load the model directly from the safetensors file
+      let vs_unet = unsafe { VarBuilder::from_mmaped_safetensors(&[unet_weights], configs.dtype, &configs.device)? };
+
+      let unet = UNet2DConditionModel::new(vs_unet, in_channels, 4, use_flash_attn, unet_config)?;
+
+      let unet = UNetModel::new_with_inner(unet);
+
+      let unet = Arc::new(unet);
+
+      model_cache.set_unet(unet.clone())?;
+
+      unet
+    },
+  };
+
+  let init_latent_dist: DiagonalGaussianDistribution = vae.encode(&input_image)?;
+
+  // Parse and calculate the img2img strength
+  let img2img_strength = match i2i_strength {
+    None => 0.75f64,
+    Some(strength) => (strength as f64) / 100.0f64,
+  };
+
+  // Create LCM guidance scale embeddings
+  let embedding_dim = 1280;
+  let guidance_scale_embedding = scheduler.get_guidance_scale_embedding(guidance_scale, embedding_dim, &configs.device, configs.dtype)?;
+
+  // Generate latents from input image using the LCM approach
+  println!("Generating latents from input image...");
+  let init_latents = (init_latent_dist.sample()? * vae_scale)?;
+  println!("Initial latents shape: {:?}", init_latents.shape());
+
+  // Calculate timesteps for LCM with proper img2img handling
+  let timesteps = LCMScheduler::get_timesteps_for_steps(configs.scheduler_steps, img2img_strength);
+  let t_start = timesteps[0]; // First denoising step
+
+  // Add noise at the appropriate timestep for img2img
+  println!("Adding noise to latents for t_start={}", t_start);
+  let noise = init_latents.randn_like(0f64, 1f64)?;
+  let latents = scheduler.add_noise(&init_latents, noise, t_start)?;
+
+  println!("Latents initialized successfully");
+  println!("Initial noise shape: {:?}", latents.shape());
+
+  println!("Starting diffusion process...");
+
+  let mut latents = latents;
+  let timesteps: Vec<_> = scheduler.timesteps().iter().copied().collect();
+
+  for (timestep_index, &timestep) in timesteps.iter().enumerate() {
+    println!("Processing step {}/{} @ timestamp = {}", timestep_index + 1, timesteps.len(), timestep);
+
+    let latent_model_input = if use_guide_scale { Tensor::cat(&[&latents, &latents], 0)? } else { latents.clone() };
+
+    let latent_model_input = scheduler.scale_model_input(latent_model_input, timestep);
+
+    // Use the guidance scale embedding in the UNet inference with the proper method name
+    let mut noise_pred = match unet.forward_with_guidance(&latent_model_input, timestep as f64, &text_embeddings, &guidance_scale_embedding) {
+      Ok(pred) => pred,
+      Err(e) => {
+        println!("UNet inference failed with error: {}", e);
+        return Err(anyhow::anyhow!("UNet inference failed: {}", e));
+      },
+    };
+
+    if use_guide_scale {
+      let chunks = noise_pred.chunk(2, 0)?;
+      let (noise_pred_uncond, noise_pred_text) = (&chunks[0], &chunks[1]);
+      noise_pred = (noise_pred_uncond + (guidance_scale * (noise_pred_text - noise_pred_uncond)?)?)?;
+    }
+
+    // Apply the LCM scheduler step
+    latents = scheduler.step(&noise_pred, timestep, &latents, timestep_index, configs.scheduler_steps)?;
+  }
+
+  println!("Diffusion process completed, decoding image...");
+  let image = vae.decode(&(latents / vae_scale)?)?;
+
+  println!("VAE decode completed");
+
+  println!("Scaling image back");
+  let image = ((image / 2.)? + 0.5)?;
+
+  println!("Normalized image values");
+  let image = (image.clamp(0f32, 1.)? * 255.)?;
+
+  println!("Convert to int8");
+  let image = image.to_dtype(DType::U8)?;
+
+  println!("Converted to 8-bit format");
+
+  let image = tensor_to_image_buffer(&image.i(0)?)?;
+
+  Ok(image)
+}

crates/desktop/src-tauri/src/ml/stable_diffusion/mod.rs

@@ -1,3 +1,4 @@
 pub mod get_vae_scale;
 pub mod infer_clip_text_embeddings;
-pub mod stable_diffusion_pipeline;
+pub mod stable_diffusion_pipeline;
+pub mod lcm_pipeline;

crates/desktop/src-tauri/src/ml/stable_diffusion/stable_diffusion_pipeline.rs

@@ -1,19 +1,24 @@
-use std::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard};
+use std::sync::{Arc, RwLock, RwLockReadGuard, RwLockWriteGuard};
 
+use crate::events::notification_event::{NotificationModelType, NotificationEvent};
 use crate::ml::image::dynamic_image_to_tensor::dynamic_image_to_tensor;
 use crate::ml::image::tensor_to_image_buffer::{tensor_to_image_buffer, RgbImage};
 use crate::ml::model_cache::ModelCache;
 use crate::ml::model_file::StableDiffusionVersion;
+use crate::ml::model_type::ModelType;
+use crate::ml::models::unet_model::UNetModel;
 use crate::ml::prompt_cache::PromptCache;
 use crate::ml::stable_diffusion::get_vae_scale::get_vae_scale;
 use crate::ml::stable_diffusion::infer_clip_text_embeddings::infer_clip_text_embeddings;
 use crate::state::app_config::AppConfig;
 use anyhow::{anyhow, Error as E, Result};
 use candle_core::{DType, IndexOp, Tensor, D};
-use candle_transformers::models::stable_diffusion::vae::DiagonalGaussianDistribution;
+use candle_transformers::models::stable_diffusion::vae::{AutoEncoderKL, DiagonalGaussianDistribution};
 use image::DynamicImage;
 use log::info;
 use rand::Rng;
+use tauri::{AppHandle, Emitter};
+use crate::state::app_dir::AppDataRoot;
 
 pub struct Args<'a> {
     pub image: &'a DynamicImage,
@@ -24,31 +29,48 @@ pub struct Args<'a> {
     pub configs: &'a AppConfig,
     pub model_cache: &'a ModelCache,
     pub prompt_cache: &'a PromptCache,
+    pub app_data_root: &'a AppDataRoot,
+    pub app: &'a AppHandle,
 }
 
 pub fn stable_diffusion_pipeline(args: Args<'_>) -> Result<RgbImage> {
+    let Args { 
+        prompt, 
+        uncond_prompt, 
+        cfg_scale, 
+        i2i_strength, 
+        configs, 
+        model_cache, 
+        prompt_cache, 
+        app, 
+        image,
+        app_data_root,
+    } = args;
+    
+    let weights_dir = app_data_root.weights_dir();
+    
     println!("Starting image generation with the following configuration:");
-    println!("  Model: {:?}", args.configs.sd_version);
-    println!("  Prompt: {}", args.prompt);
-    println!("  Steps: {}", args.configs.scheduler_steps);
-    println!("  Device: {:?}", args.configs.device);
+    println!("  Model: {:?}", configs.sd_version);
+    println!("  Prompt: {}", prompt);
+    println!("  Steps: {}", configs.scheduler_steps);
+    println!("  Device: {:?}", configs.device);
 
-    println!("Model dimensions: {}x{}", args.configs.sd_config.width, args.configs.sd_config.height);
+    println!("Model dimensions: {}x{}", configs.sd_config.width, configs.sd_config.height);
 
     // TODO(bt,2025-02-18): The scheduler is `EulerAncestralDiscreteScheduler`, but we may want to port an LCM scheduler.
     //  This is a target for performance improvement
     //  See: https://github.com/huggingface/candle/issues/1331
-    let mut scheduler = args.configs.sd_config.build_scheduler(
-        args.configs.scheduler_steps)?;
+    let mut scheduler = configs.sd_config.build_scheduler(
+        configs.scheduler_steps)?;
 
-    let seed = args.configs.seed.unwrap_or_else(|| rand::thread_rng().gen());
-    args.configs.device.set_seed(seed)?;
+    let seed = configs.seed.unwrap_or_else(|| rand::thread_rng().gen());
+    configs.device.set_seed(seed)?;
 
-    println!("Using seed: {}", seed);
+    info!("Using seed: {}", seed);
 
-    let guidance_scale = match args.cfg_scale {
+    let guidance_scale = match cfg_scale {
         Some(guidance_scale) => guidance_scale,
-        None => match args.configs.sd_version {
+        None => match configs.sd_version {
             StableDiffusionVersion::V1_5
             | StableDiffusionVersion::V2_1
             | StableDiffusionVersion::Xl => 7.5,
@@ -59,29 +81,31 @@ pub fn stable_diffusion_pipeline(args: Args<'_>) -> Result<RgbImage> {
 
     let use_guide_scale = guidance_scale > 1.0;
 
-    println!("Using guide scale: {}", use_guide_scale);
+    info!("Using guide scale: {}", use_guide_scale);
 
     info!("Checking if prompt is cached");
-    let maybe_cached = args.prompt_cache.get_copy(&args.prompt)?;
+    let maybe_cached = prompt_cache.get_copy(&prompt)?;
 
     let mut text_embeddings = if let Some(tensor) = maybe_cached {
         tensor
     } else {
+        
         info!("Prompt is NOT cached! Calculating embedding...");
         let tensor = infer_clip_text_embeddings(
-            &args.prompt,
-            &args.uncond_prompt,
+            &prompt,
+            &uncond_prompt,
             None, // tokenizer
             None, // clip_weights
             None, // clip2_weights
-            args.configs.sd_version,
-            &args.configs.sd_config,
+            configs.sd_version,
+            &configs.sd_config,
             false, // use_f16
-            &args.configs.device,
-            args.configs.dtype,
+            &configs.device,
+            configs.dtype,
             use_guide_scale,
+            weights_dir,
         )?;
-        args.prompt_cache.store_copy(&args.prompt, &tensor)?;
+        prompt_cache.store_copy(&prompt, &tensor)?;
         tensor
     };
 
@@ -90,36 +114,111 @@ pub fn stable_diffusion_pipeline(args: Args<'_>) -> Result<RgbImage> {
     println!("Loading input image into tensor...");
 
     let input_image = dynamic_image_to_tensor(
-        args.image,
-        &args.configs.device,
-        args.configs.dtype)?;
+        image,
+        &configs.device,
+        configs.dtype)?;
 
     println!("Reference image shape: {:?}", input_image.shape());
 
-    let vae_scale = get_vae_scale(args.configs.sd_version);
+    let vae_scale = get_vae_scale(configs.sd_version);
 
-    let init_latent_dist : DiagonalGaussianDistribution = args.model_cache.vae_encode(&input_image)?;
+    let maybe_vae = model_cache.get_vae()?;
+    
+    let vae = match maybe_vae {
+        Some(vae) => vae,
+        None => {
+            info!("No vae found in cache; loading...");
+            
+            let mut notify_download_complete = false;
+            //if !vae_file.exists() {
+            if true {
+                notify_download_complete = true;
+                app.emit("notification", NotificationEvent::ModelDownloadStarted {
+                    model_name: "sdxl-turbo-vae",
+                    model_type: NotificationModelType::Vae,
+                })?;
+            }
+            
+            let vae_file = weights_dir.model_path(&ModelType::SdxlTurboVae);
+            
+            let vae = configs
+              .sd_config
+              .build_vae(vae_file, &configs.device, configs.dtype)?;
+
+            let vae = Arc::new(vae);
+            
+            model_cache.set_vae(vae.clone())?;
+            
+            if notify_download_complete {
+                app.emit("notification", NotificationEvent::ModelDownloadComplete {
+                    model_name: "sdxl-turbo-vae",
+                    model_type: NotificationModelType::Vae,
+                })?;
+            }
+            
+            vae
+        }
+    };
+
+    let maybe_unet = model_cache.get_unet()?;
+
+    let unet = match maybe_unet {
+        Some(unet) => unet,
+        None => {
+            info!("No unet found in cache; loading...");
+
+            let mut notify_download_complete = false;
+            //if !unet_file.exists() {
+            if true {
+                notify_download_complete = true;
+                app.emit("notification", NotificationEvent::ModelDownloadStarted {
+                    model_name: "sdxl-turbo-unet",
+                    model_type: NotificationModelType::Unet,
+                })?;
+            }
+            
+            let unet_file = weights_dir.model_path(&ModelType::SdxlTurboUnet);
+
+            let unet = UNetModel::new(&configs.sd_config, unet_file, &configs.device, configs.dtype)
+              .map_err(|err| anyhow!("error initializing unet model: {:?}", err))?;
+            
+            let unet = Arc::new(unet);
+
+            model_cache.set_unet(unet.clone())?;
+
+            if notify_download_complete {
+                app.emit("notification", NotificationEvent::ModelDownloadComplete {
+                    model_name: "sdxl-turbo-unet",
+                    model_type: NotificationModelType::Unet,
+                })?;
+            }
+
+            unet
+        }
+    };
+
+    let init_latent_dist : DiagonalGaussianDistribution = vae.encode(&input_image)?;
 
     // TODO(bt,2025-02-18): This takes a little bit to generate the sample.
     //  This is a target for performance improvement
     println!("Generating latents from input image...");
     let latents = (init_latent_dist.sample()? * vae_scale)?;
     
-    //.to_device(&args.configs.device)?;
+    //.to_device(&configs.device)?;
     
     println!("Initial latents shape: {:?}", latents.shape());
 
     println!("Calculating start step for diffusion process...");
 
-    let img2img_strength = match args.i2i_strength {
+    let img2img_strength = match i2i_strength {
         None => 0.75f64,
         Some(strength) => (strength as f64) / 100.0f64,
     };
 
     let t_start = {
-        let start = args.configs.scheduler_steps - (args.configs.scheduler_steps as f64 * img2img_strength) as usize;
+        let start = configs.scheduler_steps - (configs.scheduler_steps as f64 * img2img_strength) as usize;
 
-        println!("Starting from step {} of {} (strength: {})", start, args.configs.scheduler_steps, img2img_strength);
+        println!("Starting from step {} of {} (strength: {})", start, configs.scheduler_steps, img2img_strength);
         start
     };
 
@@ -158,7 +257,7 @@ pub fn stable_diffusion_pipeline(args: Args<'_>) -> Result<RgbImage> {
 
         let latent_model_input = scheduler.scale_model_input(latent_model_input, timestep)?;
 
-        let mut noise_pred = match args.model_cache.unet_inference(&latent_model_input, timestep as f64, &text_embeddings) {
+        let mut noise_pred = match unet.inference(&latent_model_input, timestep as f64, &text_embeddings) {
             Ok(pred) => pred,
             Err(e) => {
                 println!("UNet inference failed with error: {}", e);
@@ -176,7 +275,7 @@ pub fn stable_diffusion_pipeline(args: Args<'_>) -> Result<RgbImage> {
     }
 
     println!("Diffusion process completed, decoding image...");
-    let image = args.model_cache.vae_decode(&(latents / vae_scale)?)?;
+    let image = vae.decode(&(latents / vae_scale)?)?;
     
     println!("VAE decode completed");
 

crates/desktop/src-tauri/src/state/app_config.rs

@@ -1,18 +1,19 @@
 use crate::ml::model_file::StableDiffusionVersion;
+use crate::state::app_dir::AppDataRoot;
+use crate::state::os_platform::OsPlatform;
 use crate::state::yaml_config::YamlConfig;
 use candle_core::{DType, Device};
 use candle_transformers::models::stable_diffusion::StableDiffusionConfig;
 use filesys::recursively_find_file_by_name::recursively_find_file_by_name;
 use hf_hub::api::sync::Api;
 use log::info;
+use std::path::PathBuf;
 
-const CONFIG_FILENAME : &str = "app_config.yaml";
-
-const DEFAULT_SD_IMAGE_WIDTH: usize = 512;
-const DEFAULT_SD_IMAGE_HEIGHT: usize = 512;
+const DEFAULT_SD_IMAGE_WIDTH: usize = 1024;
+const DEFAULT_SD_IMAGE_HEIGHT: usize = 1024;
 
 const DEFAULT_SD_NUMERIC_DATATYPE: DType = DType::F32;
-const SD_VERSION: StableDiffusionVersion = StableDiffusionVersion::Turbo;
+const SD_VERSION: StableDiffusionVersion = StableDiffusionVersion::V1_5;
 
 /// A central place to configure app-wide details.
 pub struct AppConfig {
@@ -37,13 +38,15 @@ pub struct AppConfig {
   
   /// Probably shouldn't be here.
   pub hf_api: Api,
+  
+  /// Root location for application data.
+  pub app_data_root: AppDataRoot,
 }
 
 impl AppConfig {
   
   pub fn init() -> anyhow::Result<Self> {
-    println!("Configuration filename: {}", CONFIG_FILENAME);
-    let yaml_configs = load_yaml_configs();
+    let yaml_configs = YamlConfig::load_from_config_file_recursive();
     
     println!("Configs: {:?}", yaml_configs);
     
@@ -52,18 +55,42 @@ impl AppConfig {
           info!("CUDA not available ({}), falling back to CPU", e);
           Device::Cpu
         });
+
     
     let sd_config = match SD_VERSION {
       StableDiffusionVersion::Turbo => {
         StableDiffusionConfig::sdxl_turbo(None, Some(DEFAULT_SD_IMAGE_HEIGHT), Some(DEFAULT_SD_IMAGE_WIDTH))
       }
-      _ => {
+      StableDiffusionVersion::V1_5 | StableDiffusionVersion::V1_5Inpaint => {
+        StableDiffusionConfig::v1_5(None, Some(DEFAULT_SD_IMAGE_HEIGHT), Some(DEFAULT_SD_IMAGE_WIDTH))
+      }
+      StableDiffusionVersion::V2_1 | StableDiffusionVersion::V2Inpaint => {
         StableDiffusionConfig::v2_1(None, Some(DEFAULT_SD_IMAGE_HEIGHT), Some(DEFAULT_SD_IMAGE_WIDTH))
       }
+      StableDiffusionVersion::Xl | StableDiffusionVersion::XlInpaint => {
+        StableDiffusionConfig::sdxl(None, Some(DEFAULT_SD_IMAGE_HEIGHT), Some(DEFAULT_SD_IMAGE_WIDTH))
+      }
     };
     
     let hf_api = Api::new()?;
     
+    let os_platform = OsPlatform::get();
+
+    let default_data_root = match os_platform {
+      OsPlatform::Linux => yaml_configs.linux_default_data_path.as_deref(),
+      OsPlatform::MacOs => yaml_configs.mac_default_data_path.as_deref(),
+      OsPlatform::Windows => yaml_configs.windows_default_data_path.as_deref(),
+    };
+    
+    println!("Possible user-defined data root: {:?}", default_data_root);
+
+    let app_data_root = match default_data_root {
+      None => AppDataRoot::create_default()?,
+      Some(path) => AppDataRoot::create_existing(path)?,
+    };
+
+    println!("App data root: {:?}", app_data_root.path());
+
     Ok(Self {
       device,
       dtype: DEFAULT_SD_NUMERIC_DATATYPE,
@@ -71,26 +98,12 @@ impl AppConfig {
       sd_config,
       image_height: yaml_configs.image_height.unwrap_or(DEFAULT_SD_IMAGE_HEIGHT),
       image_width: yaml_configs.image_width.unwrap_or(DEFAULT_SD_IMAGE_WIDTH),
-      scheduler_steps: yaml_configs.scheduler_steps.unwrap_or(1),
+      scheduler_steps: yaml_configs.scheduler_steps.unwrap_or(4),
       scheduler_samples: yaml_configs.scheduler_samples.unwrap_or(15),
       seed: yaml_configs.seed,
       cfg_scale: yaml_configs.cfg_scale,
       hf_api,
+      app_data_root,
     })
   }
 }
-
-fn load_yaml_configs() -> YamlConfig {
-  let maybe_path = recursively_find_file_by_name(CONFIG_FILENAME, "../../../", 5)
-    .ok()
-    .flatten();
-  if let Some(path) = maybe_path {
-    let path = path.canonicalize().unwrap_or(path);
-    println!("Attempting to load configs from: {:?}", &path);
-    if let Ok(config) = YamlConfig::read_from_file(path) {
-      return config;
-    }
-  }
-  println!("Loading default configs.");
-  YamlConfig::default()
-}

crates/desktop/src-tauri/src/state/app_dir.rs

@@ -0,0 +1,186 @@
+use crate::ml::model_type::ModelType;
+use crate::state::expanduser::expanduser;
+use crate::state::os_platform::OsPlatform;
+use anyhow::anyhow;
+use directories::UserDirs;
+use std::path::{Path, PathBuf};
+use tempdir::TempDir;
+use tempfile::{Builder, NamedTempFile};
+const DEFAULT_DATA_DIR : &str = "artcraft";
+const ASSETS_SUBDIRECTORY : &str = "assets";
+const WEIGHTS_SUBDIRECTORY : &str = "weights";
+
+const TEMPORARY_SUBDIRECTORY : &str = "temp";
+
+/// The path to the application data directory, which includes "asset" and "weights" data.
+#[derive(Clone)]
+pub struct AppDataRoot {
+  path: PathBuf,
+  assets_dir: AppAssetsDir,
+  weights_dir: AppWeightsDir,
+  temp_dir: TemporaryDir,
+}
+
+#[derive(Clone)]
+pub struct AppAssetsDir {
+  path: PathBuf,
+}
+
+#[derive(Clone)]
+pub struct AppWeightsDir {
+  path: PathBuf,
+}
+
+#[derive(Clone)]
+pub struct TemporaryDir {
+  path: PathBuf,
+}
+
+impl AppDataRoot {
+  pub fn create_default() -> anyhow::Result<Self> {
+    let directory = get_default_data_dir()?;
+    Self::create_existing(directory)
+  }
+
+  pub fn create_existing<P: AsRef<Path>>(dir: P) -> anyhow::Result<Self> {
+    let mut dir = dir.as_ref().to_path_buf();
+    
+    match OsPlatform::get() {
+      OsPlatform::Linux | OsPlatform::MacOs => {
+        if let Some(d) = dir.as_os_str().to_str() {
+          dir = expanduser(d)?;
+        }
+      },
+      OsPlatform::Windows => {}
+    }
+    
+    if !dir.is_dir() {
+      println!("Creating directory {:?}", dir);
+      std::fs::create_dir_all(&dir)?;
+    }
+
+    match dir.canonicalize() {
+      Ok(d) => dir = d,
+      Err(err) => {
+        println!("Error canonicalizing {:?}: {}", dir, err);
+      }
+    }
+    
+    let assets_dir = AppAssetsDir::create_existing(dir.join(ASSETS_SUBDIRECTORY))?;
+    let weights_dir = AppWeightsDir::create_existing(dir.join(WEIGHTS_SUBDIRECTORY))?;
+    let temp_dir = TemporaryDir::create_existing(dir.join(TEMPORARY_SUBDIRECTORY))?;
+
+    Ok(Self {
+      path: dir,
+      assets_dir,
+      weights_dir,
+      temp_dir,
+    })
+  }
+  
+  pub fn assets_dir(&self) -> &AppAssetsDir {
+    &self.assets_dir
+  }
+  
+  pub fn weights_dir(&self) -> &AppWeightsDir {
+    &self.weights_dir
+  }
+
+  pub fn temp_dir(&self) -> &TemporaryDir {
+    &self.temp_dir
+  }
+
+  pub fn path(&self) -> &Path {
+    &self.path
+  }
+}
+
+impl AppAssetsDir {
+  pub fn create_existing<P: AsRef<Path>>(dir: P) -> anyhow::Result<Self> {
+    let mut dir = dir.as_ref().to_path_buf();
+    match dir.canonicalize() {
+      Ok(d) => dir = d,
+      Err(err) => {
+        println!("Error canonicalizing {:?}: {}", dir, err);
+      }
+    }
+    if !dir.exists() {
+      println!("Creating directory {:?}", dir);
+      std::fs::create_dir(&dir)?;
+    }
+    Ok(Self {
+      path: dir,
+    })
+  }
+
+  pub fn path(&self) -> &Path {
+    &self.path
+  }
+}
+
+impl AppWeightsDir {
+  pub fn create_existing<P: AsRef<Path>>(dir: P) -> anyhow::Result<Self> {
+    let mut dir = dir.as_ref().to_path_buf();
+    match dir.canonicalize() {
+      Ok(d) => dir = d,
+      Err(err) => {
+        println!("Error canonicalizing {:?}: {}", dir, err);
+      }
+    }
+    if !dir.exists() {
+      println!("Creating directory {:?}", dir);
+      std::fs::create_dir(&dir)?;
+    }
+    Ok(Self {
+      path: dir,
+    })
+  }
+
+  pub fn path(&self) -> &Path {
+    &self.path
+  }
+  
+  pub fn model_path(&self, model_registry: &ModelType) -> PathBuf {
+    self.path.join(model_registry.get_filename())
+  }
+}
+
+impl TemporaryDir {
+  pub fn create_existing<P: AsRef<Path>>(dir: P) -> anyhow::Result<Self> {
+    let mut dir = dir.as_ref().to_path_buf();
+    match dir.canonicalize() {
+      Ok(d) => dir = d,
+      Err(err) => {
+        println!("Error canonicalizing {:?}: {}", dir, err);
+      }
+    }
+    if !dir.exists() {
+      println!("Creating directory {:?}", dir);
+      std::fs::create_dir(&dir)?;
+    }
+    Ok(Self {
+      path: dir,
+    })
+  }
+
+  pub fn with_prefix(&self, prefix: &str) -> anyhow::Result<TempDir> {
+    let tempdir = TempDir::new_in(&self.path, prefix)?;
+    Ok(tempdir)
+  }
+
+  pub fn new_named_temp_file(&self) -> anyhow::Result<NamedTempFile> {
+    let tempfile = Builder::new()
+      .prefix("temp_")
+      .suffix(".bin")
+      .tempfile_in(&self.path)?;
+    Ok(tempfile)
+  }
+}
+
+// eg. /home/bob/artcraft, /Users/bob/artcraft, or C:\Users\Alice\artcraft
+fn get_default_data_dir() -> anyhow::Result<PathBuf> {
+  Ok(UserDirs::new()
+      .ok_or_else(|| anyhow!("could not determine user home directory"))?
+      .home_dir()
+      .join(DEFAULT_DATA_DIR))
+}

crates/desktop/src-tauri/src/state/expanduser.rs

@@ -0,0 +1,73 @@
+use once_cell::sync::Lazy;
+use std::io;
+use std::path::{PathBuf, MAIN_SEPARATOR};
+
+static PREFIX: Lazy<String> = Lazy::new(|| format!("~{}", MAIN_SEPARATOR));
+
+/// The "expanduser" crate doesn't compile on Windows, so we replace its functionality slightly
+pub fn expanduser<P: AsRef<str>>(path: P) -> io::Result<PathBuf> {
+
+   Ok(match path.as_ref() {
+       // matches an exact "~"
+       s if s == "~" => {
+           home_dir()?
+       },
+       // matches paths that start with `~/`
+       s if s.starts_with(&*PREFIX) => {
+           let home = home_dir()?;
+           home.join(&s[2..])
+       },
+       // // matches paths that start with `~` but not `~/`, might be a `~username/` path
+       // s if s.starts_with("~") => {
+       //     let mut parts = s[1..].splitn(2, MAIN_SEPARATOR);
+       //     let user = parts.next()
+       //         .ok_or_else(|| io::Error::new(io::ErrorKind::Other, "malformed path"))?;
+       //     let user = Passwd::from_name(&user)
+       //         .map_err(|_| io::Error::new(io::ErrorKind::Other, "error searching for user"))?
+       //         .ok_or_else(|| io::Error::new(io::ErrorKind::Other, format!("user '{}', does not exist", &user)))?;
+       //     if let Some(ref path) = parts.next() {
+       //         PathBuf::from(user.dir).join(&path)
+       //     } else {
+       //         PathBuf::from(user.dir)
+       //     }
+       // },
+       // nothing to expand, just make a PathBuf
+       s => PathBuf::from(s)
+   })
+}
+
+pub fn home_dir() -> io::Result<PathBuf> {
+    dirs::home_dir()
+        .ok_or_else(|| io::Error::new(io::ErrorKind::Other, "no home directory is set"))
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::state::expanduser::expanduser;
+    use std::env;
+    use std::path::PathBuf;
+
+    #[test]
+    #[serial_test::serial]
+    fn test_success() {
+        let old_home = env::var("HOME").expect("no home dir set");
+        let new_home = "/home/foo";
+        env::set_var("HOME", new_home);
+        let path = expanduser("~/path/to/directory");
+        env::set_var("HOME", old_home);
+        assert_eq!(path.expect("io error"), PathBuf::from("/home/foo/path/to/directory"));
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn test_only_tilde() {
+        let old_home = env::var("HOME").expect("no home dir set");
+        let new_home = "/home/foo";
+        env::set_var("HOME", new_home);
+        let pathstr = "~";
+        let path = expanduser(pathstr);
+        env::set_var("HOME", old_home);
+        assert_eq!(path.expect("io error"), PathBuf::from("/home/foo"));
+    }
+
+}

crates/desktop/src-tauri/src/state/mod.rs

@@ -1,2 +1,5 @@
+pub mod app_config;
+pub mod app_dir;
+pub mod os_platform;
 pub mod yaml_config;
-pub mod app_config;
+pub mod expanduser;

crates/desktop/src-tauri/src/state/os_platform.rs

@@ -0,0 +1,20 @@
+
+#[derive(Copy,Clone,Eq,PartialEq,Debug)]
+pub enum OsPlatform {
+  Linux,
+  MacOs,
+  Windows,
+}
+
+impl OsPlatform {
+  pub fn get() -> OsPlatform {
+    // Totally scientific.
+    // https://doc.rust-lang.org/std/env/consts/constant.OS.html
+    match std::env::consts::OS {
+      "linux" => OsPlatform::Linux,
+      "windows" => OsPlatform::Windows,
+      "macos" | "apple" => OsPlatform::MacOs,
+      _ => OsPlatform::Linux,
+    }
+  }
+}

crates/desktop/src-tauri/src/state/yaml_config.rs

@@ -1,7 +1,10 @@
+use filesys::recursively_find_file_by_name::recursively_find_file_by_name;
 use std::fs::File;
 use std::io::BufReader;
 use std::path::Path;
 
+const CONFIG_FILENAME : &str = "app_config.yaml";
+
 /// Configuration read or derived from a YAML config file at startup
 /// This is not meant to contain any runtime config values.
 /// This may not be exhaustive.
@@ -13,6 +16,15 @@ pub struct YamlConfig {
   pub scheduler_samples: Option<usize>,
   pub seed: Option<u64>,
   pub cfg_scale: Option<f64>,
+
+  /// Override the default location to store data on Linux
+  pub linux_default_data_path: Option<String>,
+
+  /// Override the default location to store data on Windows
+  pub windows_default_data_path: Option<String>,
+  
+  /// Override the default location to store data on Mac
+  pub mac_default_data_path: Option<String>,
 }
 
 impl YamlConfig {
@@ -22,4 +34,19 @@ impl YamlConfig {
     let config = serde_yaml::from_reader(reader)?;
     Ok(config)
   }
+
+  pub fn load_from_config_file_recursive() -> YamlConfig {
+    let maybe_path = recursively_find_file_by_name(CONFIG_FILENAME, "../../../", 5)
+      .ok()
+      .flatten();
+    if let Some(path) = maybe_path {
+      let path = path.canonicalize().unwrap_or(path);
+      println!("Attempting to load configs from: {:?}", &path);
+      if let Ok(config) = YamlConfig::read_from_file(path) {
+        return config;
+      }
+    }
+    println!("Couldn't load config file. Using default configs.");
+    YamlConfig::default()
+  }
 }

crates/desktop/src-tauri/src/threads/downloader_thread.rs

@@ -0,0 +1,172 @@
+use crate::events::notification_event::{NotificationEvent, NotificationModelType};
+use crate::ml::model_type::ModelType;
+use crate::state::app_dir::AppDataRoot;
+use crate::transfer::download::{download_async, ProgressUpdate};
+use futures::stream::FuturesUnordered;
+use futures::StreamExt;
+use log::{debug, error, info};
+use std::collections::VecDeque;
+use std::path::PathBuf;
+use std::sync::mpsc::Receiver;
+use std::sync::{Arc, Mutex};
+use tauri::{AppHandle, Emitter};
+use tempfile::NamedTempFile;
+use tokio::task::JoinHandle;
+
+const NOTIFICATION_CHANNEL_NAME : &str = "notification";
+const MAX_FILES : usize = 8;
+const CHUNK_SIZE : usize = 1024 * 1024;
+const PARALLEL_FAILURES : usize = 8;
+const MAX_RETRIES : usize = 30;
+
+struct ModelDownloadStatusChannel {
+  pub model_name: String,
+  pub model_type: NotificationModelType,
+  pub final_file_path: PathBuf,
+  pub receiver: Receiver<ProgressUpdate>,
+}
+
+pub async fn downloader_thread(app_data_root: AppDataRoot, app: AppHandle) -> ! {
+
+  let mut status_check_queue : Arc<Mutex<Vec<ModelDownloadStatusChannel>>> = Arc::new(Mutex::new(Vec::new()));
+
+  {
+    let status_check_queue2 = status_check_queue.clone();
+    let app2 = app.clone();
+
+    tokio::spawn(download_notify_thread(status_check_queue2, app2));
+  }
+
+  let mut download_queue = VecDeque::new();
+
+  // TODO: Automatic enqueue of known important models + enqueue new models on-demand
+  download_queue.push_back(ModelType::ClipJson);
+  //download_queue.push_back(ModelType::SdxlTurboUnet);
+  //download_queue.push_back(ModelType::SdxlTurboVae);
+  download_queue.push_back(ModelType::SdxlTurboClipEncoder); // TODO(bt): Why is this still needed?
+  //download_queue.push_back(ModelType::SdxlTurboClipEncoder2);
+  download_queue.push_back(ModelType::SimianLuoLcmDreamshaperV7Unet);
+  download_queue.push_back(ModelType::LykonDreamshaper7Vae);
+  download_queue.push_back(ModelType::LykonDreamshaper7TextEncoderFp16);
+
+  let mut handles = FuturesUnordered::new();
+
+  while let Some(model) = download_queue.pop_front() {
+    let url = model.get_download_url().to_string();
+
+    let filename = app_data_root.weights_dir().model_path(&model);
+
+    if filename.exists() {
+      info!("File already exists: {:?}", &filename);
+      continue;
+    }
+
+    let headers = None;
+    
+    let (tx, rx) = std::sync::mpsc::channel();
+
+    info!("Download: {:?}", filename);
+
+    let temp_file = match app_data_root.temp_dir().new_named_temp_file() {
+      Ok(file) => file,
+      Err(err) => {
+        error!("Couldn't create a temporary file: {:?}", err);
+        continue;
+      }
+    };
+
+    let result = app.emit(NOTIFICATION_CHANNEL_NAME, NotificationEvent::ModelDownloadStarted {
+      model_name: model.get_name(),
+      model_type: model.get_notification_type(),
+    });
+
+    if let Err(err) = result {
+      error!("Could not emit event: {}", err);
+    }
+
+    let task = tokio::spawn(download_async(
+      url,
+      temp_file,
+      filename.clone(),
+      MAX_FILES,
+      CHUNK_SIZE,
+      PARALLEL_FAILURES,
+      MAX_RETRIES,
+      headers,
+      Some(tx),
+    ));
+
+    match status_check_queue.lock() {
+      Err(err) => error!("Could not unlock: {:?}", err),
+      Ok(mut lock) => {
+        lock.push(ModelDownloadStatusChannel {
+          receiver: rx,
+          final_file_path: filename.clone(),
+          model_name: model.get_name().to_string(),
+          model_type: model.get_notification_type(),
+        })
+      }
+    }
+
+    handles.push(task);
+  }
+
+  while let Some(result) = handles.next().await {
+    match result {
+      Err(err) => error!("Error downloading: {:?}", err),
+      Ok(_) => {},
+    }
+  }
+
+  loop {
+    // TODO(bt,2025-03-10): Queue management for new files entering the queue.
+    tokio::time::sleep(std::time::Duration::from_millis(60_000)).await;
+  }
+}
+
+pub async fn download_notify_thread(queue: Arc<Mutex<Vec<ModelDownloadStatusChannel>>>, app: AppHandle) -> ! {
+  loop {
+    match queue.lock() {
+      Err(err) => error!("Could not unlock: {:?}", err),
+      Ok(mut lock) => {
+        debug!(">>>>> Notifier has {} items", lock.len());
+
+        let mut reenqueue = Vec::with_capacity(lock.len());
+
+        for item in (*lock).drain(..) {
+          if let Ok(update) = item.receiver.try_recv() {
+            let result = app.emit(NOTIFICATION_CHANNEL_NAME, NotificationEvent::ModelDownloadProgress {
+              model_name: &item.model_name,
+              model_type: item.model_type,
+              currently_downloaded_bytes: update.complete,
+              total_file_bytes: update.total_length,
+            });
+
+            if let Err(err) = result {
+              error!("Could not emit event: {}", err);
+            }
+          }
+
+          if item.final_file_path.exists() {
+            debug!("No longer sending notifications for {:?}", &item.final_file_path);
+
+            let result = app.emit(NOTIFICATION_CHANNEL_NAME, NotificationEvent::ModelDownloadComplete {
+              model_name: &item.model_name,
+              model_type: item.model_type,
+            });
+
+            if let Err(err) = result {
+              error!("Could not emit event: {}", err);
+            }
+          } else {
+            reenqueue.push(item);
+          }
+        }
+
+        lock.extend(reenqueue);
+      }
+    }
+
+    tokio::time::sleep(std::time::Duration::from_millis(500)).await;
+  }
+}

crates/desktop/src-tauri/src/threads/mod.rs

@@ -0,0 +1 @@
+pub mod downloader_thread;

crates/desktop/src-tauri/src/transfer/download.rs

@@ -0,0 +1,241 @@
+//! Implemented from https://github.com/huggingface/hf_transfer/blob/main/src/lib.rs
+
+use crate::transfer::error::Error;
+use crate::transfer::util::{exponential_backoff, BASE_WAIT_TIME, MAX_WAIT_TIME};
+use futures::stream::FuturesUnordered;
+use futures::StreamExt;
+use reqwest::header::{
+  HeaderMap, HeaderName, HeaderValue, AUTHORIZATION, CONTENT_RANGE,
+  RANGE,
+};
+use reqwest::Url;
+use std::collections::HashMap;
+use std::fmt::Display;
+use std::fs::remove_file;
+use std::io::SeekFrom;
+use std::path::Path;
+use std::sync::Arc;
+use std::sync::atomic::{AtomicUsize, Ordering};
+use std::sync::mpsc::Sender;
+use std::time::Duration;
+use log::info;
+use tempfile::NamedTempFile;
+use tokio::fs::OpenOptions;
+use tokio::io::AsyncWriteExt;
+use tokio::io::AsyncSeekExt;
+use tokio::sync::Semaphore;
+use tokio::time::sleep;
+
+pub struct ProgressUpdate {
+  pub complete: usize,
+  pub total_length: usize,
+}
+
+/// max_files: Number of open file handles, which determines the maximum number of parallel downloads
+/// parallel_failures:  Number of maximum failures of different chunks in parallel (cannot exceed max_files)
+/// max_retries: Number of maximum attempts per chunk. (Retries are exponentially backed off + jitter)
+///
+/// The number of threads can be tuned by the environment variable `TOKIO_WORKER_THREADS` as documented in
+/// https://docs.rs/tokio/latest/tokio/runtime/struct.Builder.html#method.worker_threads
+#[allow(clippy::too_many_arguments)]
+pub async fn download_async<P: AsRef<Path>>(
+  url: String,
+  filename: NamedTempFile,
+  final_filename: P,
+  max_files: usize,
+  chunk_size: usize,
+  parallel_failures: usize,
+  max_retries: usize,
+  input_headers: Option<HashMap<String, String>>,
+  maybe_progress_sender: Option<Sender<ProgressUpdate>>,
+) -> Result<(), Error> {
+  info!(">>> Downloading: {}", filename.as_ref().display());
+
+  let client = reqwest::Client::builder()
+    // https://github.com/hyperium/hyper/issues/2136#issuecomment-589488526
+    .http2_keep_alive_timeout(Duration::from_secs(15))
+    .build()?;
+
+  let mut headers = HeaderMap::new();
+  let mut auth_token = None;
+  if let Some(input_headers) = input_headers {
+    headers.reserve(input_headers.len());
+    for (k, v) in input_headers {
+      let name: HeaderName = k
+        .try_into()
+        .map_err(|err| Error::new_err(format!("Invalid header: {err}")))?;
+      let value: HeaderValue = AsRef::<str>::as_ref(&v)
+        .try_into()
+        .map_err(|err| Error::new_err(format!("Invalid header value: {err}")))?;
+      if name == AUTHORIZATION {
+        auth_token = Some(value);
+      } else {
+        headers.insert(name, value);
+      }
+    }
+  };
+
+  let response = if let Some(token) = auth_token.as_ref() {
+    client.get(&url).header(AUTHORIZATION, token)
+  } else {
+    client.get(&url)
+  }
+    .headers(headers.clone())
+    .header(RANGE, "bytes=0-0")
+    .send()
+    .await
+    .map_err(|err| Error::new_err(format!("Error while downloading: {err}")))?
+    .error_for_status()
+    .map_err(|err| Error::new_err(err.to_string()))?;
+
+  // Only call the final redirect URL to avoid overloading the Hub with requests and also
+  // altering the download count
+  let redirected_url = response.url();
+  if Url::parse(&url)
+    .map_err(|err| Error::new_err(format!("failed to parse url: {err}")))?
+    .host()
+    == redirected_url.host()
+  {
+    if let Some(token) = auth_token {
+      headers.insert(AUTHORIZATION, token);
+    }
+  }
+
+  let content_range = response
+    .headers()
+    .get(CONTENT_RANGE)
+    .ok_or(Error::new_err("No content length".to_string()))?
+    .to_str()
+    .map_err(|err| Error::new_err(format!("Error while downloading: {err}")))?;
+
+  let size: Vec<&str> = content_range.split('/').collect();
+  // Content-Range: bytes 0-0/702517648
+  // https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Range
+  let length: usize = size
+    .last()
+    .ok_or(Error::new_err(
+      "Error while downloading: No size was detected".to_string(),
+    ))?
+    .parse()
+    .map_err(|err| Error::new_err(format!("Error while downloading: {err}")))?;
+
+
+  info!("File {:?} is {:?} length", filename.as_ref(), length);
+
+  let mut handles = FuturesUnordered::new();
+  let semaphore = Arc::new(Semaphore::new(max_files));
+  let parallel_failures_semaphore = Arc::new(Semaphore::new(parallel_failures));
+  let max_chunk_complete = Arc::new(AtomicUsize::new(0));
+
+  for start in (0..length).step_by(chunk_size) {
+    let url = redirected_url.to_string();
+    let filename = filename.as_ref().to_owned();
+    let client = client.clone();
+    let headers = headers.clone();
+
+    let stop = std::cmp::min(start + chunk_size - 1, length);
+    let semaphore = semaphore.clone();
+    let parallel_failures_semaphore = parallel_failures_semaphore.clone();
+    let maybe_progress = maybe_progress_sender.clone();
+    let max_chunk_complete = max_chunk_complete.clone();
+    handles.push(tokio::spawn(async move {
+      let permit = semaphore
+        .acquire_owned()
+        .await
+        .map_err(|err| Error::new_err(format!("Error while downloading: {err}")))?;
+      let mut chunk = download_chunk(&client, &url, &filename, start, stop, headers.clone()).await;
+      let mut i = 0;
+      if parallel_failures > 0 {
+        while let Err(dlerr) = chunk {
+          if i >= max_retries {
+            return Err(Error::new_err(format!(
+              "Failed after too many retries ({max_retries}): {dlerr}"
+            )));
+          }
+          let parallel_failure_permit = parallel_failures_semaphore.clone().try_acquire_owned().map_err(|err| {
+            Error::new_err(format!(
+              "Failed too many failures in parallel ({parallel_failures}): {dlerr} ({err})"
+            ))
+          })?;
+
+          let wait_time = exponential_backoff(BASE_WAIT_TIME, i, MAX_WAIT_TIME);
+          sleep(Duration::from_millis(wait_time as u64)).await;
+
+          chunk = download_chunk(&client, &url, &filename, start, stop, headers.clone()).await;
+          i += 1;
+          drop(parallel_failure_permit);
+          
+          if let Some(sender) = maybe_progress.as_ref() {
+            // NB: Calling fetch_max twice. First call will return the lower value.
+            let _old = max_chunk_complete.fetch_max(stop, Ordering::SeqCst);
+            let max_chunk = max_chunk_complete.fetch_max(stop, Ordering::SeqCst);
+            let progress_update = ProgressUpdate { 
+              complete: max_chunk,
+              total_length: length,
+            };
+            if let Err(_err) = sender.send(progress_update) {
+              // Fail silently, not essential to issue progress updates
+            }
+          }
+        }
+      }
+      drop(permit);
+      chunk.map_err(|e| Error::new_err(format!("Downloading error {e}"))).and(Ok(stop - start))
+    }));
+  }
+
+  // Output the chained result
+  while let Some(result) = handles.next().await {
+    match result {
+      Ok(Ok(size)) => {
+        //if let Some(ref callback) = callback {
+        //  callback.call((size,), None)?;
+        //}
+      }
+      Ok(Err(py_err)) => {
+        return Err(py_err);
+      }
+      Err(err) => {
+        return Err(Error::new_err(format!(
+          "Error while downloading: {err}"
+        )));
+      }
+    }
+  }
+
+  let file = filename.persist(final_filename)
+    .map_err(|e| Error::new_err(format!("Could not rename to final filename: {:?}", e)))?;
+
+  file.sync_all()?;
+
+  Ok(())
+}
+
+async fn download_chunk<P: AsRef<Path>>(
+  client: &reqwest::Client,
+  url: &str,
+  filename: P,
+  start: usize,
+  stop: usize,
+  headers: HeaderMap,
+) -> Result<(), Error> {
+  // Process each socket concurrently.
+  let range = format!("bytes={start}-{stop}");
+  let mut file = OpenOptions::new()
+    .write(true)
+    .truncate(false)
+    .create(true)
+    .open(filename)
+    .await?;
+  file.seek(SeekFrom::Start(start as u64)).await?;
+  let response = client
+    .get(url)
+    .headers(headers)
+    .header(RANGE, range)
+    .send()
+    .await?
+    .error_for_status()?;
+  let content = response.bytes().await?;
+  file.write_all(&content).await?;
+  Ok(())
+}

crates/desktop/src-tauri/src/transfer/error.rs

@@ -0,0 +1,49 @@
+//! Implemented from https://github.com/huggingface/hf_transfer/blob/main/src/lib.rs
+
+use std::fmt::Display;
+use reqwest::header::ToStrError;
+
+#[derive(Debug)]
+pub enum Error {
+  Io(std::io::Error),
+  Request(reqwest::Error),
+  ToStrError(ToStrError),
+  Misc(String),
+}
+
+impl From<std::io::Error> for Error {
+  fn from(value: std::io::Error) -> Self {
+    Self::Io(value)
+  }
+}
+
+impl From<reqwest::Error> for Error {
+  fn from(value: reqwest::Error) -> Self {
+    Self::Request(value)
+  }
+}
+
+impl From<ToStrError> for Error {
+  fn from(value: ToStrError) -> Self {
+    Self::ToStrError(value)
+  }
+}
+
+impl Display for Error {
+  fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+    match self {
+      Self::Io(io) => write!(f, "Io: {io}"),
+      Self::Request(req) => write!(f, "Request: {req}"),
+      Self::ToStrError(req) => write!(f, "Response non ascii: {req}"),
+      Self::Misc(err) => write!(f, "Misc error: {err}"),
+    }
+  }
+}
+
+impl std::error::Error for Error {}
+
+impl Error {
+  pub fn new_err(reason: String) -> Self {
+    Self::Misc(reason)
+  }
+}

crates/desktop/src-tauri/src/transfer/mod.rs

@@ -0,0 +1,4 @@
+pub mod download;
+pub mod upload;
+pub mod util;
+pub mod error;

crates/desktop/src-tauri/src/transfer/upload.rs

@@ -0,0 +1,189 @@
+//! Implemented from https://github.com/huggingface/hf_transfer/blob/main/src/lib.rs
+
+use crate::transfer::error::Error;
+use crate::transfer::util::{exponential_backoff, BASE_WAIT_TIME, MAX_WAIT_TIME};
+use futures::stream::FuturesUnordered;
+use futures::StreamExt;
+use reqwest::header::CONTENT_LENGTH;
+use std::collections::HashMap;
+use std::fmt::Display;
+use std::io::SeekFrom;
+use std::sync::Arc;
+use std::time::Duration;
+use tokio::fs::OpenOptions;
+use tokio::io::{AsyncReadExt, AsyncSeekExt};
+use tokio::sync::Semaphore;
+use tokio::time::sleep;
+use tokio_util::codec::{BytesCodec, FramedRead};
+
+/// parts_urls: Dictionary consisting of part numbers as keys and the associated url as values
+/// completion_url: The url that should be called when the upload is finished
+/// max_files: Number of open file handles, which determines the maximum number of parallel uploads
+/// parallel_failures:  Number of maximum failures of different chunks in parallel (cannot exceed max_files)
+/// max_retries: Number of maximum attempts per chunk. (Retries are exponentially backed off + jitter)
+///
+/// The number of threads can be tuned by the environment variable `TOKIO_WORKER_THREADS` as documented in
+/// https://docs.rs/tokio/latest/tokio/runtime/struct.Builder.html#method.worker_threads
+///
+/// See https://docs.aws.amazon.com/AmazonS3/latest/userguide/mpuoverview.html for more information
+/// on the multipart upload
+#[allow(clippy::too_many_arguments)]
+fn multipart_upload(
+  file_path: String,
+  parts_urls: Vec<String>,
+  chunk_size: u64,
+  max_files: usize,
+  parallel_failures: usize,
+  max_retries: usize,
+) -> Result<Vec<HashMap<String, String>>, Error> {
+  if parallel_failures > max_files {
+    return Err(Error::new_err(
+      "Error parallel_failures cannot be > max_files".to_string(),
+    ));
+  }
+  if (parallel_failures == 0) != (max_retries == 0) {
+    return Err(Error::new_err(
+      "For retry mechanism you need to set both `parallel_failures` and `max_retries`"
+        .to_string(),
+    ));
+  }
+
+  tokio::runtime::Builder::new_multi_thread()
+    .enable_all()
+    .build()?
+    .block_on(async {
+      upload_async(
+        file_path,
+        parts_urls,
+        chunk_size,
+        max_files,
+        parallel_failures,
+        max_retries,
+      )
+        .await
+    })
+}
+
+
+
+#[allow(clippy::too_many_arguments)]
+async fn upload_async(
+  file_path: String,
+  parts_urls: Vec<String>,
+  chunk_size: u64,
+  max_files: usize,
+  parallel_failures: usize,
+  max_retries: usize,
+) -> Result<Vec<HashMap<String, String>>, Error> {
+  let client = reqwest::Client::new();
+
+  let mut handles = FuturesUnordered::new();
+  let semaphore = Arc::new(Semaphore::new(max_files));
+  let parallel_failures_semaphore = Arc::new(Semaphore::new(parallel_failures));
+
+  for (part_number, part_url) in parts_urls.iter().enumerate() {
+    let url = part_url.to_string();
+    let path = file_path.to_owned();
+    let client = client.clone();
+
+    let start = (part_number as u64) * chunk_size;
+    let semaphore = semaphore.clone();
+    let parallel_failures_semaphore = parallel_failures_semaphore.clone();
+    handles.push(tokio::spawn(async move {
+      let permit = semaphore
+        .clone()
+        .acquire_owned()
+        .await
+        .map_err(|err| Error::new_err(format!("Error acquiring semaphore: {err}")))?;
+      let mut chunk = upload_chunk(&client, &url, &path, start, chunk_size).await;
+      let mut i = 0;
+      if parallel_failures > 0 {
+        while let Err(ul_err) = chunk {
+          if i >= max_retries {
+            return Err(Error::new_err(format!(
+              "Failed after too many retries ({max_retries}): {ul_err}"
+            )));
+          }
+
+          let parallel_failure_permit = parallel_failures_semaphore.clone().try_acquire_owned().map_err(|err| {
+            Error::new_err(format!(
+              "Failed too many failures in parallel ({parallel_failures}): {ul_err} ({err})"
+            ))
+          })?;
+
+          let wait_time = exponential_backoff(BASE_WAIT_TIME, i, MAX_WAIT_TIME);
+          sleep(Duration::from_millis(wait_time as u64)).await;
+
+          chunk = upload_chunk(&client, &url, &path, start, chunk_size).await;
+          i += 1;
+          drop(parallel_failure_permit);
+        }
+      }
+      drop(permit);
+      chunk.map_err(|e|{
+        //match e {
+        //  Error::Io(io) => Error::new_err(format!("Io error {io}")),
+        //  Error::Request(req) => Error::new_err(format!("Error while sending chunk {req}")),
+        //  Error::ToStrError(req) => Error::new_err(format!("Response header contains non ASCII chars: {req}")),
+        //}
+        e
+      }).map(|chunk| (part_number, chunk, chunk_size))
+    }));
+  }
+
+  let mut results: Vec<HashMap<String, String>> = vec![HashMap::default(); parts_urls.len()];
+
+  while let Some(result) = handles.next().await {
+    match result {
+      Ok(Ok((part_number, headers, size))) => {
+        //if let Some(ref callback) = callback {
+        //  callback.call((size,), None)?;
+        //}
+        results[part_number] = headers;
+      }
+      Ok(Err(py_err)) => {
+        return Err(py_err);
+      }
+      Err(err) => {
+        return Err(Error::new_err(format!(
+          "Error occurred while uploading: {err}"
+        )));
+      }
+    }
+  }
+
+  Ok(results)
+}
+
+async fn upload_chunk(
+  client: &reqwest::Client,
+  url: &str,
+  path: &str,
+  start: u64,
+  chunk_size: u64,
+) -> Result<HashMap<String, String>, Error> {
+  let mut options = OpenOptions::new();
+  let mut file = options.read(true).open(path).await?;
+  let file_size = file.metadata().await?.len();
+  let bytes_transferred = std::cmp::min(file_size - start, chunk_size);
+
+  file.seek(SeekFrom::Start(start)).await?;
+  let chunk = file.take(chunk_size);
+
+  let response = client
+    .put(url)
+    .header(CONTENT_LENGTH, bytes_transferred)
+    .body(reqwest::Body::wrap_stream(FramedRead::new(
+      chunk,
+      BytesCodec::new(),
+    )))
+    .send()
+    .await?;
+  let response = response.error_for_status()?;
+  let mut headers = HashMap::new();
+  for (name, value) in response.headers().into_iter() {
+    headers.insert(name.to_string(), value.to_str()?.to_owned());
+  }
+  Ok(headers)
+}
+

crates/desktop/src-tauri/src/transfer/util.rs

@@ -0,0 +1,15 @@
+//! Implemented from https://github.com/huggingface/hf_transfer/blob/main/src/lib.rs
+
+use rand::{thread_rng, Rng};
+use std::fmt::Display;
+
+pub const BASE_WAIT_TIME: usize = 300;
+pub const MAX_WAIT_TIME: usize = 10_000;
+
+fn jitter() -> usize {
+  thread_rng().gen_range(0..=500)
+}
+
+pub fn exponential_backoff(base_wait_time: usize, n: usize, max: usize) -> usize {
+  (base_wait_time + n.pow(2) + jitter()).min(max)
+}

crates/desktop/src-tauri/tauri.conf.json

@@ -1,6 +1,6 @@
 {
   "$schema": "https://schema.tauri.app/config/2.0.0-rc",
-  "productName": "Realtime Desktop App",
+  "productName": "ArtCraft",
   "version": "0.1.0",
   "identifier": "com.unique.value.thing.why",
   "build": {
@@ -12,9 +12,9 @@
   "app": {
     "windows": [
       {
-        "title": "Realtime Desktop App Title Goes Here",
-        "width": 1200,
-        "height": 800,
+        "title": "ArtCraft",
+        "width": 2400,
+        "height": 1300,
         "resizable": true,
         "fullscreen": false
       }

crates/desktop/src-tauri/tests/test_dynamic_library_loading.rs

@@ -0,0 +1,7 @@
+
+#[test]
+fn test_dynamic_library_loading() {
+  let name = libloading::library_filename("cublas");
+
+  assert_eq!(name, "libcublas.so");
+}

crates/service/web/storyteller_web/src/configs/plans/plan_list.rs

@@ -118,24 +118,40 @@ pub static DEVELOPMENT_PREMIUM_PLANS: Lazy<HashSet<Plan>> = Lazy::new(|| {
 
 /// All paid-for production premium plans in the system.
 pub static PRODUCTION_PREMIUM_PLANS: Lazy<HashSet<Plan>> = Lazy::new(|| {
-    let mut plans = HashSet::new();
+  let mut plans = HashSet::new();
 
-    // ========== All-locale plans ==========
+  // ========== All-locale plans ==========
 
-    plans.insert(PlanBuilder::new("fakeyou_plus")
-        .plan_category(PlanCategory::Paid)
-        .stripe_product_id("prod_MoLv23HrxPiY7D")
-        .stripe_price_id("price_1M4jDCEU5se17MekOaJ92HYX")
-        .cost_per_month_dollars(7)
-        .tts_base_priority_level(20)
-        .tts_max_character_length(2048)
-        .tts_max_duration_seconds(30)
-        .web_vc_base_priority_level(20)
-        .web_vc_requires_frontend_keepalive(false)
-        .lipsync_requires_frontend_keepalive(false)
-        .build());
+  // Old "plus" price
+  plans.insert(PlanBuilder::new("fakeyou_plus")
+      .plan_category(PlanCategory::Paid)
+      .stripe_product_id("prod_MoLv23HrxPiY7D")
+      .stripe_price_id("price_1M4jDCEU5se17MekOaJ92HYX")
+      .cost_per_month_dollars(7)
+      .tts_base_priority_level(20)
+      .tts_max_character_length(2048)
+      .tts_max_duration_seconds(30)
+      .web_vc_base_priority_level(20)
+      .web_vc_requires_frontend_keepalive(false)
+      .lipsync_requires_frontend_keepalive(false)
+      .build());
 
-    plans.insert(PlanBuilder::new("fakeyou_pro")
+  // New "plus" price
+  plans.insert(PlanBuilder::new("fakeyou_plus_plan")
+      .plan_category(PlanCategory::Paid)
+      .stripe_product_id("prod_RvC5qUgLTSLGSD")
+      .stripe_price_id("price_1R1Lh8EU5se17MekaUNtWc59")
+      .cost_per_month_dollars(12)
+      .tts_base_priority_level(20)
+      .tts_max_character_length(2048)
+      .tts_max_duration_seconds(30)
+      .web_vc_base_priority_level(20)
+      .web_vc_requires_frontend_keepalive(false)
+      .lipsync_requires_frontend_keepalive(false)
+      .build());
+
+  // Old "pro" price
+  plans.insert(PlanBuilder::new("fakeyou_pro")
         .plan_category(PlanCategory::Paid)
         .stripe_product_id("prod_MoLw8nA6eFxHzc")
         .stripe_price_id("price_1M4jEQEU5se17MeksNfA0EKm")
@@ -149,7 +165,23 @@ pub static PRODUCTION_PREMIUM_PLANS: Lazy<HashSet<Plan>> = Lazy::new(|| {
         .can_remove_visual_watermarks(true)
         .build());
 
-    plans.insert(PlanBuilder::new("fakeyou_elite")
+  // New "pro" price
+  plans.insert(PlanBuilder::new("fakeyou_pro_plan")
+      .plan_category(PlanCategory::Paid)
+      .stripe_product_id("prod_RvC6nB9rw6yAsI")
+      .stripe_price_id("price_1R1LhzEU5se17Mek3JLwN3Gg")
+      .cost_per_month_dollars(25)
+      .tts_base_priority_level(30)
+      .tts_max_character_length(3072)
+      .tts_max_duration_seconds(60 * 2)
+      .web_vc_base_priority_level(30)
+      .web_vc_requires_frontend_keepalive(false)
+      .lipsync_requires_frontend_keepalive(false)
+      .can_remove_visual_watermarks(true)
+      .build());
+
+  // Old "elite" price
+  plans.insert(PlanBuilder::new("fakeyou_elite")
         .plan_category(PlanCategory::Paid)
         .stripe_product_id("prod_MoLxQmLA4R24fv")
         .stripe_price_id("price_1M4jFREU5se17Mekc8pQaSKB")
@@ -163,6 +195,21 @@ pub static PRODUCTION_PREMIUM_PLANS: Lazy<HashSet<Plan>> = Lazy::new(|| {
         .can_remove_visual_watermarks(true)
         .build());
 
+  // New "elite" price
+  plans.insert(PlanBuilder::new("fakeyou_elite_plan")
+      .plan_category(PlanCategory::Paid)
+      .stripe_product_id("prod_RvC6i84YiAYStd")
+      .stripe_price_id("price_1R1LiQEU5se17MekiJdBcsHQ")
+      .cost_per_month_dollars(40)
+      .tts_base_priority_level(40)
+      .tts_max_character_length(4096)
+      .tts_max_duration_seconds(60 * 5)
+      .web_vc_base_priority_level(40)
+      .web_vc_requires_frontend_keepalive(false)
+      .lipsync_requires_frontend_keepalive(false)
+      .can_remove_visual_watermarks(true)
+      .build());
+
     // ========== Spanish plans ==========
 
     plans.insert(PlanBuilder::new("fakeyou_basic")
@@ -293,7 +340,7 @@ mod test {
     // making ourselves check twice when we make additions, removals, or other changes.
     #[test]
     fn test_number_of_plans_is_expected() {
-        assert_eq!(11, ALL_PLANS_BY_SLUG.len());
+        assert_eq!(14, ALL_PLANS_BY_SLUG.len());
     }
 
     #[test]
@@ -303,7 +350,7 @@ mod test {
             .filter(|plan| !plan.is_synthetic_plan()) // NB: Synthetic plans are production!
             .collect::<Vec<&Plan>>();
 
-        let expected = 4;
+        let expected = 7;
 
         assert_eq!(expected, production_plans.len());
         assert_eq!(expected, PRODUCTION_PREMIUM_PLANS.len());
@@ -346,6 +393,9 @@ mod test {
             "prod_MoLw8nA6eFxHzc".to_string(), // NB: Production
             "prod_MoLxQmLA4R24fv".to_string(), // NB: Production
             "prod_MoLyt8qMDmscjr".to_string(), // NB: Production
+            "prod_RvC5qUgLTSLGSD".to_string(), // NB: Production - new "plus" product
+            "prod_RvC6nB9rw6yAsI".to_string(), // NB: Production - new "pro" product
+            "prod_RvC6i84YiAYStd".to_string(), // NB: Production - new "elite" product
         ];
 
         assert_that(&product_ids).contains_all_of(&expected_product_ids.iter());
@@ -366,6 +416,9 @@ mod test {
             "price_1M4jEQEU5se17MeksNfA0EKm".to_string(), // NB: Production
             "price_1M4jFREU5se17Mekc8pQaSKB".to_string(), // NB: Production
             "price_1M4jGPEU5se17Mek2FiztNE5".to_string(), // NB: Production
+            "price_1R1Lh8EU5se17MekaUNtWc59".to_string(), // NB: Production - new "plus" price
+            "price_1R1LhzEU5se17Mek3JLwN3Gg".to_string(), // NB: Production - new "pro" price
+            "price_1R1LiQEU5se17MekiJdBcsHQ".to_string(), // NB: Production - new "elite" price
         ];
 
         assert_that(&price_ids).contains_all_of(&expected_price_ids.iter());

crates/vendor/candle/candle-examples/Cargo.toml

@@ -1,59 +1,52 @@
 [package]
 name = "candle-examples"
-version.workspace = true
-edition.workspace = true
-description.workspace = true
-repository.workspace = true
-keywords.workspace = true
-categories.workspace = true
-license.workspace = true
-readme = "README.md"
+edition = "2021"
 
 [dependencies]
-accelerate-src = { workspace = true, optional = true }
-candle = { workspace = true }
-candle-datasets = { workspace = true, optional = true }
-candle-nn = { workspace = true }
-candle-transformers = { workspace = true }
-candle-flash-attn = { workspace = true, optional = true }
-candle-onnx = { workspace = true, optional = true }
+accelerate-src = { version = "0.3.2", optional = true }
+candle = { path = "../candle-core", package = "candle-core" }
+candle-datasets = { path = "../candle-datasets", optional = true }
+candle-nn = { path = "../candle-nn", package = "candle-nn" }
+candle-transformers = { path = "../candle-transformers" }
+candle-flash-attn = { path = "../candle-flash-attn", optional = true }
+candle-onnx = { path = "../candle-onnx", optional = true }
 
 csv = "1.3.0"
-cudarc = { workspace = true, optional = true }
-half = { workspace = true, optional = true }
-hf-hub = { workspace = true, features = ["tokio"] }
-image = { workspace = true }
-intel-mkl-src = { workspace = true, optional = true }
-num-traits = { workspace = true }
+cudarc = { version = "0.13.5", features = ["std", "cublas", "cublaslt", "curand", "driver", "nvrtc", "f16", "cuda-version-from-build-system", "dynamic-linking"], default-features=false, optional = true }
+half = { version = "2.3.1", features = ["num-traits", "use-intrinsics", "rand_distr"], optional = true }
+hf-hub = { version = "0.4.1", features = ["tokio"] }
+image = { version = "0.25.2", default-features = false, features = ["jpeg", "png"] }
+intel-mkl-src = { version = "0.8.1", features = ["mkl-static-lp64-iomp"], optional = true }
+num-traits = "0.2.15"
 palette = { version = "0.7.6", optional = true }
 enterpolation = { version = "0.2.1", optional = true}
 pyo3 = { version = "0.22.0", features = ["auto-initialize", "abi3-py311"], optional = true }
-rayon = { workspace = true }
+rayon = "1.7.0"
 rubato = { version = "0.15.0", optional = true }
-safetensors = { workspace = true }
-serde = { workspace = true }
-serde_json = { workspace = true }
+safetensors = "0.4.1"
+serde = { version = "1.0.171", features = ["derive"] }
+serde_json = "1.0.99"
 symphonia = { version = "0.5.3", features = ["all"], optional = true }
-tokenizers = { workspace = true, features = ["onig"] }
+tokenizers = { version = "0.19.1", default-features = false, features = ["onig"] }
 cpal = { version = "0.15.2", optional = true }
 pdf2image = { version = "0.1.2" , optional = true}
 
 [dev-dependencies]
-anyhow = { workspace = true }
-byteorder = { workspace = true }
-clap = { workspace = true }
-imageproc = { workspace = true }
-memmap2 = { workspace = true }
-rand = { workspace = true }
-ab_glyph = { workspace = true }
-tracing = { workspace = true }
-tracing-chrome = { workspace = true }
-tracing-subscriber = { workspace = true }
+anyhow = { version = "1", features = ["backtrace"] }
+byteorder = "1.4.3"
+clap = { version = "4.2.4", features = ["derive"] }
+imageproc = { version = "0.24.0", default-features = false }
+memmap2 = { version = "0.9.3", features = ["stable_deref_trait"] }
+rand = "0.8.5"
+ab_glyph = "0.2.23"
+tracing = "0.1.37"
+tracing-chrome = "0.7.1"
+tracing-subscriber = "0.3.7"
 # Necessary to disambiguate with tokio in wasm examples which are 1.28.1
 tokio = "1.43.0"
 
 [build-dependencies]
-anyhow = { workspace = true }
+anyhow = { version = "1", features = ["backtrace"] }
 bindgen_cuda = { version = "0.1.1", optional = true }
 
 [features]
@@ -122,3 +115,7 @@ required-features = ["onnx"]
 [[example]]
 name = "colpali"
 required-features = ["pdf2image"]
+
+[[example]]
+name = "lcm_debug_test"
+path = "examples/lcm/debug_test.rs"

crates/vendor/candle/candle-flash-attn/build.rs

@@ -86,6 +86,8 @@ fn main() -> Result<()> {
         .arg("--expt-relaxed-constexpr")
         .arg("--expt-extended-lambda")
         .arg("--use_fast_math")
+        .arg("--compiler-options")
+        .arg("-fPIC")
         .arg("--verbose");
 
     if let Ok(target) = std::env::var("TARGET") {
@@ -100,7 +102,12 @@ fn main() -> Result<()> {
     println!("cargo:rustc-link-search={}", build_dir.display());
     println!("cargo:rustc-link-lib=flashattention");
     println!("cargo:rustc-link-lib=dylib=cudart");
-    println!("cargo:rustc-link-lib=dylib=stdc++");
+
+
+    // TODO(bt,2025-03-12): Fix builds on windows
+    //println!("cargo:rustc-link-lib=dylib=stdc++");
+
+
 
     Ok(())
 }

crates/vendor/candle/candle-flash-attn/cutlass/.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,23 @@
+---
+name: Bug report
+about: Create a bug report to help us improve CUTLASS
+title: "[BUG]"
+labels: "? - Needs Triage, bug"
+assignees: ''
+
+---
+
+**Describe the bug**
+A clear and concise description of what the bug is.
+
+**Steps/Code to reproduce bug**
+Follow this guide http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports to craft a minimal bug report. This helps us reproduce the issue you're having and resolve the issue more quickly.
+
+**Expected behavior**
+A clear and concise description of what you expected to happen.
+
+**Environment details (please complete the following information):**
+ - Environment location: [Bare-metal, Docker, Cloud(specify cloud provider)] 
+
+**Additional context**
+Add any other context about the problem here.

crates/vendor/candle/candle-flash-attn/cutlass/.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,5 @@
+blank_issues_enabled: true
+contact_links:
+  - name: CUTLASS Discord
+    url: https://discord.gg/nvidiadeveloper
+    about: Come chat about using and contributing to CUTLASS!

crates/vendor/candle/candle-flash-attn/cutlass/.github/ISSUE_TEMPLATE/documentation_request.md

@@ -0,0 +1,35 @@
+---
+name: Documentation request
+about: Report incorrect or needed documentation to improve CUTLASS
+title: "[DOC]"
+labels: "? - Needs Triage, documentation"
+assignees: ''
+
+---
+
+## Report incorrect documentation
+
+**Location of incorrect documentation**
+Provide links and line numbers if applicable.
+
+**Describe the problems or issues found in the documentation**
+A clear and concise description of what you found to be incorrect.
+
+**Steps taken to verify documentation is incorrect**
+List any steps you have taken:
+
+**Suggested fix for documentation**
+Detail proposed changes to fix the documentation if you have any.
+
+---
+
+## Report needed documentation
+
+**Report needed documentation**
+A clear and concise description of what documentation you believe it is needed and why.
+
+**Describe the documentation you'd like**
+A clear and concise description of what you want to happen.
+
+**Steps taken to search for needed documentation**
+List any steps you have taken:

crates/vendor/candle/candle-flash-attn/cutlass/.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,20 @@
+---
+name: Feature request
+about: Suggest an idea for CUTLASS
+title: "[FEA]"
+labels: "? - Needs Triage, feature request"
+assignees: ''
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I wish I could use CUTLASS to do [...]
+
+**Describe the solution you'd like**
+A clear and concise description of what you want to happen.
+
+**Describe alternatives you've considered**
+A clear and concise description of any alternative solutions or features you've considered.
+
+**Additional context**
+Add any other context, code examples, or references to existing implementations about the feature request here.

crates/vendor/candle/candle-flash-attn/cutlass/.github/ISSUE_TEMPLATE/submit_question.md

@@ -0,0 +1,10 @@
+---
+name: Submit question
+about: Ask a general question about CUTLASS
+title: "[QST]"
+labels: "? - Needs Triage, question"
+assignees: ''
+
+---
+
+**What is your question?**

crates/vendor/candle/candle-flash-attn/cutlass/.github/workflows/labeler.yml

@@ -0,0 +1,11 @@
+name: "Pull Request Labeler"
+on:
+- pull_request_target
+
+jobs:
+  triage:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/labeler@main
+      with:
+        repo-token: "${{ secrets.GITHUB_TOKEN }}"

crates/vendor/candle/candle-flash-attn/cutlass/.github/workflows/new-issues-to-triage-projects.yml

@@ -0,0 +1,35 @@
+name: Auto Assign New Issues to Triage Project
+
+on:
+  issues:
+    types: [opened]
+
+env:
+  GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+jobs:
+  assign_one_project:
+    runs-on: ubuntu-latest
+    name: Assign to New Issues to Triage Project
+    steps:
+    - name: Process bug issues
+      uses: docker://takanabe/github-actions-automate-projects:v0.0.1
+      if: contains(github.event.issue.labels.*.name, 'bug') && contains(github.event.issue.labels.*.name, '? - Needs Triage')
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        GITHUB_PROJECT_URL: https://github.com/NVIDIA/cutlass
+        GITHUB_PROJECT_COLUMN_NAME: 'Needs prioritizing'
+    - name: Process feature issues
+      uses: docker://takanabe/github-actions-automate-projects:v0.0.1
+      if: contains(github.event.issue.labels.*.name, 'feature request') && contains(github.event.issue.labels.*.name, '? - Needs Triage')
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        GITHUB_PROJECT_URL: https://github.com/NVIDIA/cutlass
+        GITHUB_PROJECT_COLUMN_NAME: 'Needs prioritizing'
+    - name: Process other issues
+      uses: docker://takanabe/github-actions-automate-projects:v0.0.1
+      if: contains(github.event.issue.labels.*.name, '? - Needs Triage') && (!contains(github.event.issue.labels.*.name, 'bug') && !contains(github.event.issue.labels.*.name, 'feature request'))
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        GITHUB_PROJECT_URL: https://github.com/NVIDIA/cutlass
+        GITHUB_PROJECT_COLUMN_NAME: 'Needs prioritizing'

crates/vendor/candle/candle-flash-attn/cutlass/.github/workflows/stale.yml

@@ -0,0 +1,57 @@
+name: Mark inactive issues and pull requests
+
+on:
+  schedule:
+    - cron: "0 * * * *"
+
+jobs:
+  mark-inactive-30d:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Mark 30 day inactive issues and pull requests
+        uses: actions/stale@v3
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          stale-issue-message: >
+            This issue has been labeled `inactive-30d` due to no recent activity in the past 30 days.
+            Please close this issue if no further response or action is needed.
+            Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.
+            This issue will be labeled `inactive-90d` if there is no activity in the next 60 days.
+          stale-issue-label: "inactive-30d"
+          exempt-issue-labels: "0 - Blocked,0 - Backlog,good first issue"
+          days-before-issue-stale: 30
+          days-before-issue-close: -1
+          stale-pr-message: >
+            This PR has been labeled `inactive-30d` due to no recent activity in the past 30 days.
+            Please close this PR if it is no longer required.
+            Otherwise, please respond with a comment indicating any updates.
+            This PR will be labeled `inactive-90d` if there is no activity in the next 60 days.
+          stale-pr-label: "inactive-30d"
+          exempt-pr-labels: "0 - Blocked,0 - Backlog,good first issue"
+          days-before-pr-stale: 30
+          days-before-pr-close: -1
+          operations-per-run: 50
+  mark-inactive-90d:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Mark 90 day inactive issues and pull requests
+        uses: actions/stale@v3
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          stale-issue-message: >
+            This issue has been labeled `inactive-90d` due to no recent activity in the past 90 days.
+            Please close this issue if no further response or action is needed.
+            Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.
+          stale-issue-label: "inactive-90d"
+          exempt-issue-labels: "0 - Blocked,0 - Backlog,good first issue"
+          days-before-issue-stale: 90
+          days-before-issue-close: -1
+          stale-pr-message: >
+            This PR has been labeled `inactive-90d` due to no recent activity in the past 90 days.
+            Please close this PR if it is no longer required.
+            Otherwise, please respond with a comment indicating any updates.
+          stale-pr-label: "inactive-90d"
+          exempt-pr-labels: "0 - Blocked,0 - Backlog,good first issue"
+          days-before-pr-stale: 90
+          days-before-pr-close: -1
+          operations-per-run: 50

crates/vendor/candle/candle-flash-attn/cutlass/.gitignore

@@ -0,0 +1,3 @@
+# PyCache files
+__pycache__/
+cutlass_library.egg-info/

crates/vendor/candle/candle-flash-attn/cutlass/CHANGELOG.md

@@ -0,0 +1,467 @@
+# NVIDIA CUTLASS Changelog
+
+## [3.6.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.6.0) (2024-10-03)
+
+- [Hopper structured sparse GEMM](./examples/62_hopper_sparse_gemm/62_hopper_sparse_gemm.cu).
+  + [FP16](./test/unit/gemm/device/sm90_sparse_gemm_f16_f16_f32_tensor_op_f32.cu)
+  + [FP8](./test/unit/gemm/device/sm90_sparse_gemm_f8_f8_f32_tensor_op_f32.cu)
+  + [INT8](./test/unit/gemm/device/sm90_sparse_gemm_s8_s8_s32_tensor_op_s32.cu)
+  + [TF32](./test/unit/gemm/device/sm90_sparse_gemm_tf32_tf32_f32_tensor_op_f32.cu)
+- A refactor to the CUTLASS 3.x convolution `kernel::ConvUniversal` [API](./include/cutlass/conv/kernel/sm90_implicit_gemm_tma_warpspecialized.hpp) to bring it in line with `gemm::GemmUniversal`. Now the 3.x convolution API is no longer considered as a beta API.
+- [An improved mixed input GEMM](./examples/55_hopper_mixed_dtype_gemm/README.md) and a [lookup table implementation](./examples/55_hopper_mixed_dtype_gemm/55_hopper_int4_fp8_gemm.cu) for `INT4`x`FP8` scale-only mode.
+- [EVT nodes for Top-K selection and softmax](./include/cutlass/epilogue/fusion/sm90_visitor_topk_softmax.hpp) and [GEMM example using those](./examples/61_hopper_gemm_with_topk_and_softmax/61_hopper_gemm_with_topk_and_softmax.cu).
+- [Programmatic Dependent Launch](./include/cutlass/arch/grid_dependency_control.h) (PDL) that leverages a new Hopper feature to speedup two back-to-back kernels, and its corresponding [documentations](./media/docs/dependent_kernel_launch.md).
+- [A new debugging tool, synclog](./include/cutlass/arch/synclog.hpp), for dumping out all synchronization events from within a kernel to a file. Please see [synclog documentation](./media/docs/utilities.md#debugging-asynchronous-kernels-with-cutlasss-built-in-synclog-tool) for details.
+- A new TMA-enabled [epilogue](./include/cutlass/epilogue/collective/sm90_epilogue_array_tma_warpspecialized.hpp) for grouped GEMM that brings significant performance improvement, as well as its EVT support.
+- A SIMT-enabled pointer-array [epilogue](./include/cutlass/epilogue/collective/sm70_epilogue_vectorized_array.hpp).
+- A new [Ping-Pong kernel schedule for Grouped GEMM](./include/cutlass/gemm/kernel/sm90_gemm_array_tma_warpspecialized_pingpong.hpp) and some other optimizations.
+- [A new instantiation strategy for CUTLASS profiler kernels](./python/cutlass_library/sm90_shapes.py) along with [improved documentation for instantiation level in CUTLASS profiler](./media/docs/profiler.md#instantiating-more-kernels-with-hopper).
+- A new hardware support for comparisons and computations of [`cutlass::bfloat16_t`](./include/cutlass/bfloat16.h)
+- Fixed use of isnan on Windows for [`half_t`](./test/unit/core/functional.cu).
+  Various improvements and fixed from the community and CUTLASS team. Thanks to everyone who submitted PRs!
+
+## [3.5.1](https://github.com/NVIDIA/cutlass/releases/tag/v3.5.1) (2024-07-25)
+
+- [Minimal SM90 WGMMA + TMA GEMM example in 100 lines of code](./examples/cute/tutorial/wgmma_sm90.cu)
+- [Exposure of L2 `cache_hint`s in TMA copy atoms](./include/cute/arch/copy_sm90_tma.hpp#L48)
+- Exposure of raster order and tile swizzle extent in [CUTLASS library profiler](./media/docs/profiler.md#GEMM), and
+[example 48](./examples/48_hopper_warp_specialized_gemm/48_hopper_warp_specialized_gemm.cu).
+- [TMA store based and EVT supported epilogues](./include/cutlass/epilogue/collective/sm90_epilogue_array_tma_warpspecialized.hpp) for [Hopper pointer array batched kernels](./test/unit/gemm/device/sm90_gemm_f16_f16_f16_tensor_op_f32_ptr_array.cu).
+- A new [`GemmSparseUniversal` API for CUTLASS 2.x Ampere kernels](./include/cutlass/gemm/device/gemm_sparse_universal.h) to enable serial and parallel split-k for sparse tensor cores and new tiny tile sizes to better support LLM inferrence:
+  + [FP16 TN](./test/unit/gemm/device/gemm_f16t_f16n_f32t_tensor_op_f32_sparse_sm80.cu#L269-L393) and [NT](./test/unit/gemm/device/gemm_f16n_f16t_f32t_tensor_op_f32_sparse_sm80.cu#L269-L411).
+  + [int8 TN](./test/unit/gemm/device/gemm_s8t_s8n_s32t_tensor_op_s32_sparse_sm80.cu#L264-L452).
+  + [int4 TN](./test/unit/gemm/device/gemm_s4t_s4n_s32t_tensor_op_s32_sparse_sm80.cu#L264-L452).
+  + [FP32 TN](./test/unit/gemm/device/gemm_f32t_f32n_f32t_tensor_op_f32_sparse_sm80.cu#L427-L642) and [NT](./test/unit/gemm/device/gemm_f32n_f32t_f32t_tensor_op_f32_sparse_sm80.cu#L427-L456).
+- [CUDA host adapter](./include/cutlass/cuda_host_adapter.hpp) extensions to support TMA descriptor construction driver APIs.
+- Inclusion of more [Hopper fprop, dgrad, and wgrad convolution kernels in CUTLASS library and profiler](./python/cutlass_library/generator.py).
+- Support for residual add (beta != 0) in convolution kernels.
+- A new convolution [epilogue](./examples/16_ampere_tensorop_conv2dfprop/ampere_tensorop_conv2dfprop.cu#L269) for CUTLASS 2.x to support non-packed NHWC output.
+- A refactor of [include files throughout CUTLASS core directories](./include/cutlass/gemm/collective/collective_mma_decl.hpp) to reduce circular dependencies and [tests to guard against them](./test/self_contained_includes/CMakeLists.txt).
+- [A guide for setting up VSCode to work well with CUTLASS](./media/docs/ide_setup.md) and [expanded code style guide](./media/docs/programming_guidelines.md).
+- Better support for MSVC as a host compiler.
+- Many performance optimizations, improvements, and bug fixes including fixes for FlashAttention-2.
+- Optimal code generation with CUDA toolkit versions 12.4 and 12.5u1.
+
+## [3.5.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.5.0) (2024-04-09)
+
+- Implicit GEMM Convolutions targeting Hopper SM90A via WGMMA + [TMA im2col](./include/cute/atom/copy_traits_sm90_im2col.hpp)
+  + Native implementation in CUTLASS 3.x using CuTe, mirroring the [same design hierarchy as that of GEMMs](./media/docs/gemm_api_3x.md).
+  + Support for 1D, 2D, and 3D convolutions in a [rank-agnostic fashion](./include/cutlass/conv/convnd_problem_shape.hpp).
+  + Support for [Fprop](./test/unit/conv/device_3x/fprop/sm90_conv3d_fprop_implicit_gemm_s8_s8_s32_tensorop_s32.cu), [Dgrad](./test/unit/conv/device_3x/dgrad/sm90_conv2d_dgrad_implicit_gemm_f16_f16_f32_tensorop_f16.cu), and [Wgrad](./test/unit/conv/device_3x/wgrad/sm90_conv1d_wgrad_implicit_gemm_f16_f16_f32_tensorop_f16.cu) algorithms
+  + [CUTLASS profiler support](./python/cutlass_library/conv3x_emitter.py) for 2D and 3D convolutions implemented via the 3.x API.
+  + NOTE: this is a beta release. Further updates to CUTLASS will include major performance improvements, feature enablement, and possible breaking changes to the API until 3.7 release. Your feedback is welcome on the design!
+- Support for [Ada (SM89) FP8 tensor cores via the 2.x API](./examples/58_ada_fp8_gemm/ada_fp8_gemm.cu). Requires CUDA 12.4 or newer.
+- [Ampere gather/scatter convolution example](./examples/59_ampere_gather_scatter_gemm/README.md) in CuTe and CUTLASS 3.x
+  + Showcasing how custom kernels can be written and optimized using CUTLASS 3.x and CuTe and the general strategy for implementing convolutions as specializations of GETTs.
+  + Implementation of a coarse grained sparse gather/scatter kernel achieving peak performance on Ampere class tensor cores.
+- 32x and 16x tile sizes are added to CUTLASS 2.x to improve the performance of narrow-tall and wide-short matrices.
+  + [Ampere FP16 TN](./test/unit/gemm/device/gemm_f16t_f16n_f16t_tensor_op_f32_sm80.cu) and [NT](./test/unit/gemm/device/gemm_f16n_f16t_f16t_tensor_op_f32_sm80.cu#L227-L301), [Ampere INT8 TN](./test/unit/gemm/device/gemm_s8t_s8n_s8t_tensor_op_s32_sm80.cu#L392-L1342), [Ampere INT4 TN](./test/unit/gemm/device/gemm_s4t_s4n_s4t_tensor_op_s32_sm80.cu#L372-L934).
+  + [Turing FP16 TN](./test/unit/gemm/device/gemm_f16t_f16n_f16t_tensor_op_f32_sm75.cu#L55-L394), [Turing INT8 TN](./test/unit/gemm/device/gemm_s8t_s8n_s8t_tensor_op_s32_sm75.cu#L166-L537), [Turing INT4 TN](./test/unit/gemm/device/gemm_s4t_s4n_s4t_tensor_op_s32_sm75.cu#L310-L564).
+- Updates to CuTe documentation for [`cute::Tensor<>`](./media/docs/cute/03_tensor.md), [MMA atoms](./media/docs/cute/0t_mma_atom.md), and an overhauled [CuTe GEMM tutorial series](./examples/cute/tutorial).
+- Extensions to CuTe to support [L2 prefetching](./include/cute/algorithm/prefetch.hpp) and [TMA store+reductions](./include/cute/arch/copy_sm90_tma.hpp#L1337).
+- Remove C++11 requirement on a few CUTLASS 2.x API header files. All CUTLASS files now require C++17.
+- Fixes to greatly reduce build warnings.
+- Updates and bugfixes from the community (thanks!)
+
+## [3.4.1](https://github.com/NVIDIA/cutlass/releases/tag/v3.4.1) (2024-02-14)
+
+- Statically available [CUTLASS Version macros](./include/cutlass/version.h) that allow for handling API changes between CUTLASS releases on the users' side.
+- Improvements for Hopper [Group-GEMMs](./examples/57_hopper_grouped_gemm) and [Pointer-Array Batched GEMMs](./examples/56_hopper_ptr_array_batched_gemm).
+- Updates and bugfixes from the community (thanks!).
+
+## [3.4.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.4.0) (2024-01-12)
+* Expanded [Mixed-input Hopper GEMMs](./examples/55_hopper_mixed_dtype_gemm) support covering {16-bit, 8-bit} x {8-bit, 4-bit} input types with fast numerical converters and group scaling factors.
+* Performance improvements to [Mixed-input Hopper GEMMs](./examples/55_hopper_mixed_dtype_gemm)
+* Beta release of [Pointer-Array Batched GEMMs](./examples/56_hopper_ptr_array_batched_gemm) now available on Hopper GPUs utilizing TMA and WGMMA (requires CUDA 12.3 or above).
+* Beta release of [Group-GEMM](./examples/57_hopper_grouped_gemm) utilizing TMA and WGMMA (requires CUDA 12.3 or above).
+* [Ampere Sparse GEMM](./examples/15_ampere_sparse_tensorop_gemm/ampere_sparse_tensorop_gemm_with_visitor.cu) supports Epilogue Visitor Tree (EVT) now.
+* NamedBarriers usability improvement and list of [ReservedNamedBarriers](./include/cutlass/arch/barrier.h) has been officially released.
+* Improved [CuTe documentation](./media/docs/cute/) including improved clarity and depth of [Quickstart](./media/docs/cute/00_quickstart.md), [CuTe Layout](./media/docs/cute/01_layout.md), and [CuTe Layout Algebra](./media/docs/cute/02_layout_algebra.md). Associated code comments, post-conditions, and details in [CuTe Core Unit Tests](./test/unit/cute/core/) also improved.
+
+## [3.3](https://github.com/NVIDIA/cutlass/releases/tag/v3.3.0) (2023-10-31)
+* [Mixed-input Hopper GEMMs](./examples/55_hopper_mixed_dtype_gemm) support covering 16-bit x 8-bit input operand types.
+* [Mixed-input Ampere GEMMs](https://github.com/NVIDIA/cutlass/pull/1084) with support for canonical layouts (TN). The implementation supports upcast on operandB {fp16, bf16} x {s8, u8}, and upcast on operandA {s8, u8} x {fp16, bf16}.
+* [Copy Async based Hopper GEMMs](./test/unit/gemm/device/sm90_gemm_bf16_bf16_bf16_alignx_tensor_op_f32_warpspecialized_cooperative.cu) - which support lower than 16B aligned input tensors.
+* Kernel schedules and Builder support for mixed precision and Copy Async GEMMs with < 16B aligned input tensors.  
+* Profiler support for lower-aligned Hopper GEMMs.
+* Performance Improvements to [Scatter-Gather Hopper Example](./examples/52_hopper_gather_scatter_fusion).
+* Sub-Byte type fixes and improvements.
+* EVT Support for RELU with Aux bitmap tensor store (used in dRELU). See [SM90 EVT fusions](./include/cutlass/epilogue/fusion/sm90_visitor_compute_tma_warpspecialized.hpp) for details.
+* Fusion support for backprop fusions including drelu, dgelu, and dbias.
+* Support for void-C kernels and SM80 mixed-input GEMMs in the CUTLASS Python interface
+
+## [3.2.2](https://github.com/NVIDIA/cutlass/releases/tag/v3.2.2) (2023-10-25)
+* Minor patch for issue/1138
+
+## [3.2.1](https://github.com/NVIDIA/cutlass/releases/tag/v3.2.1) (2023-09-22)
+* Python support SM90 Epilogue Visitor Tree (EVT) on top of the C++ support released in 3.2.0.
+* SM80 EVT support in C++ and Python.
+* Other SM90 epilogue improvements.
+* Splitting CUTLASS library into smaller units based on operation, arch and datatypes. See [1105](https://github.com/NVIDIA/cutlass/discussions/1105) for details.
+* Making `tools/library/scripts` packageable - `tools/library/scripts` is now moving to `python/cutlass_library`. See the Python [README](./python/README.md) for details.
+* SM90 TF32 kernel improvements for all layouts.
+* SM90 rasterization direction support in the CUTLASS profiler.
+* Improvement for CUTLASS profiler build times.
+* Remove Python-C++ bindings.
+
+## [3.2.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.2.0) (2023-08-03)
+
+* New warp-specialized persistent FP8 GEMM kernel [kernel schedules](./include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_cooperative.hpp) and [mainloops](./include/cutlass/gemm/collective/sm90_mma_tma_gmma_ss_warpspecialized_fp8.hpp)  targeting Hopper architecture that achieve great performance with TMA, WGMMA, and threadblock clusters. An example showcasing [Hopper warp-specialized FP8 GEMMs](./examples/54_hopper_fp8_warp_specialized_gemm). FP8 GEMMs come with a fast accumulation mode. When enabled, problem execution might be faster but at the cost of lower accuracy because intermediate results will not periodically be promoted to a higher precision.
+* New [Epilogue Visitor Tree (EVT)](./examples/49_hopper_gemm_with_collective_builder/49_collective_builder.cu) support for Hopper TMA epilogues. EVTs allows for user-defined customized epilogue fusion patterns without having to write a new epilogue.
+* [Stream-K](./include/cutlass/gemm/kernel/sm90_tile_scheduler_stream_k.hpp) feature for Hopper. Note that this is only a functional implementation of stream-K, and should not be used for performance comparison. Optimizations are expected in a future release.
+* Improved CTA rasterization and support for CTA swizzling for Hopper kernels using the [Tile Scheduler](./include/cutlass/gemm/kernel/sm90_tile_scheduler.hpp).
+* Improved performance for [warp-specialized TensorFloat-32 (TF32) GEMM kernels](test/unit/gemm/device/sm90_gemm_tf32_tf32_f32_tensor_op_f32_gmma_rs_cluster_warpspecialized.cu) targeting Hopper TMA.
+* [Hopper GEMM+Permute](./examples/53_hopper_gemm_permute/53_hopper_gemm_permute.cu), an example of fusing tensor reordering (permutation) with GEMM mainloop or epilogue.
+* New CUTLASS 2D Convolution Python interface. New [example](./examples/python/03_basic_conv2d.ipynb) here.
+* Support for Windows (MSVC) builds. Tested with Visual Studio 2019 v16.11.27 on Windows 10.0.
+* Optimal performance using [**CUDA 12.2u1**](https://developer.nvidia.com/cuda-downloads)
+* Updates and bugfixes from the community (thanks!)
+
+## [3.1.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.1.0) (2023-04-14)
+* New CUTLASS Python interface that aims to provide an ease-of-use interface for instantiating, emitting, compiling, and running CUTLASS kernels via Python. More details [here](./python/README.md) and new [examples](./examples/python).
+* New [efficient epilogues](test/unit/gemm/device/sm90_gemm_f16_f16_f16_tensor_op_f32_cluster_warpspecialized_cooperative.cu#L783) using TMA for Hopper.
+* Support for [fused epilogues](test/unit/gemm/device/sm90_gemm_f16_f16_f16_tensor_op_f32_cluster_warpspecialized_cooperative_bias_elementwise.cu), such Bias, ReLU and GELU, using the new efficient epilogues.
+* New [warp-specialized TensorFloat-32 (TF32) GEMM kernels](test/unit/gemm/device/sm90_gemm_tf32_tf32_f32_tensor_op_f32_gmma_rs_cluster_warpspecialized.cu) targeting Hopper TMA.
+* New [*warp-specialized persistent cooperative*](./include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_cooperative.hpp) kernel design that allows for larger tile sizes and improves performance on Hopper.
+* An [example](./examples/51_hopper_gett) showcasing GEMM-Like Tensor-Tensor Contraction (GETT) capability on Hopper.
+* Epilogue builders. Similar to mainloop builders (see [example 49](./examples/49_hopper_gemm_with_collective_builder/49_collective_builder.cu)), epilogue builders aim to generate the best-possible epilogue while exposing incremental opt-ins for greater customization.
+* Profiler support for overriding kernel and epilogue builder auto schedules for 3.x API kernels, allowing specific policies to be run in the CUTLASS profiler.
+* Performance optimizations for the [*warp-specialized persistent ping-pong*](./include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_pingpong.hpp) kernel.
+* Changes to the [GEMM API 3.x](./media/docs/gemm_api_3x.md), involving the host-facing arguments and the underlying `Params` structs.
+* [FMHA Backward Pass](./examples/41_fused_multi_head_attention/fused_multi_head_attention_backward.cu) from Meta xFormers.
+* [Streamk GEMM with Broadcast](./examples/47_ampere_gemm_universal_streamk/ampere_gemm_universal_streamk_broadcast.cu) enables epilogue broadcast with StreamK GEMM.
+* [Batched B2B GEMM](./examples/13_two_tensor_op_fusion) now can run multiple Back-to-Back GEMM with the same problem size in parallel.
+* [Batched Strided GEMV](test/unit/gemm/device/gemv.cu) support both row major and column major input matrix.
+* [Permute + GEMM fusion](./examples/39_gemm_permute) can fuse Permute with following GEMM now.  Before, we only support fusing GEMM with Permute in the epilogue.
+* [Row Broadcast](./include/cutlass/epilogue/threadblock/predicated_tile_iterator_row_broadcast.h) can be fused in the epilogue.
+* The GitHub branch is renamed from `master` to `main` in this release.
+* Optimal performance using [**CUDA 12.1**](https://developer.nvidia.com/cuda-downloads)
+* Updates and bugfixes from the community (thanks!)
+
+## [3.0.0](https://github.com/NVIDIA/cutlass/releases/tag/v3.0.0) (2023-01-23)
+* [CuTe](./media/docs/cute/00_quickstart.md), a [new core library and backend](./include/cute) for CUTLASS 3.0 that defines a single Layout vocabulary type and an associated algebra of layouts for a much more expressive and composable abstraction for tensors, sets of parallel agents, and operations by said agents on tensors.
+* [A new conceptual operation hierarchy](./media/docs/cutlass_3x_design.md) that replaces the architecture-centric hierarchy of CUTLASS 2.x and [documentation for CUTLASS 3.0's GEMM API changes](./media/docs/gemm_api_3x.md).
+* Strict API backwards compatibility that exposes both 2.x and 3.x API kernels through the same [`device::GemmUniversalAdapter`](./include/cutlass/gemm/device/gemm_universal_adapter.h) and [`kernel::GemmUniversal`](./include/cutlass/gemm/kernel/gemm_universal.hpp) types, allowing users to include both APIs in the same translation units. More information can be found in the [3.x backwards compatibility section](./media/docs/cutlass_3x_backwards_compatibility.md).
+* Updates to [Functionality](./media/docs/functionality.md) which directs users on which kernels are supported via CUTLASS-2 and CUTLASS-3.
+* Updates to [Compatibility](./README.md#compatibility) Section regarding supported compilers, operating systems, CUDA Toolkits, Hardware Architectures and [Target Architecture](./README.md#Target-Architecture).
+* New warp-specialized GEMM [kernel schedules](./include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized.hpp) and [mainloops](./include/cutlass/gemm/collective/sm90_mma_tma_gmma_ss_warpspecialized.hpp) targeting Hopper architecture that achieve great performance with TMA, WGMMA, and threadblock clusters.
+* Extensions to CUTLASS profiler to support threadblock cluster shapes in library and profiler tile configurations.
+* [CUTLASS library integration](./tools/library/src/gemm_operation_3x.hpp) for 3.x API kernels built through the new `CollectiveBuilder` API, enabling CUTLASS profiler.
+* Support for [Hopper GEMMs](./examples/48_hopper_warp_specialized_gemm) through the new 3.0 API with CuTe-based exposure of the Hopper [Tensor Memory Accelerator](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#data-movement-and-conversion-instructions-cp-async-bulk-tensor) and [WGMMA Tensor Core](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#asynchronous-warpgroup-level-matrix-instructions) features.
+* Set of examples that demonstrate the usage of the new 3.0 API to easily build GEMM kernels targeting Hopper: examples [48](./examples/48_hopper_warp_specialized_gemm), [49](./examples/49_hopper_gemm_schedules_with_collective_builder), and [50](./examples/50_hopper_gemm_with_epilogue_swizzle).
+
+## [2.11.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.11.0) (2022-11-19)
+* [Stream-K](./examples/47_ampere_gemm_universal_streamk), which is a new general way to do split-K.  It can not only improve performance, but can also significantly reduce the number of tile sizes that need to be profiled to find the best one.
+* [Fused multi-head attention Kernel](./examples/41_fused_multi_head_attention).  It has two variants: one uses batched GEMM for the fixed sequence length, and the other one uses group GEMM for the variable sequence length.  Both versions just need one kernel.
+* [Dual GEMM](./examples/45_dual_gemm), which can fuse A x B and A x C into one kernel. Two GEMMs has no producer-consumer dependency.
+* Hopper improves [double precision matrix multiplication](./test/unit/gemm/device/gemm_f64n_f64t_f64t_tensor_op_f64_sm90.cu) by 2x compared to Ampere at iso-clocks. It is supported since CUDA 11.8.
+* [BLAS3](./test/unit/gemm/device/hemm_cf64_cf64_cf64_tensor_op_f64_sm90.cu) functions with Hoppers new double precision matrix multiplication instructions.
+* [ELL Block Sparse GEMM](./examples/43_ell_block_sparse_gemm), which uses an [ELL matrix](https://developer.nvidia.com/blog/accelerating-matrix-multiplication-with-block-sparse-format-and-nvidia-tensor-cores/) to describe the sparsity of A matrix.  B and output matrices are still dense. The block size can be arbitary.
+* Optimized [Group Conv](./examples/42_ampere_tensorop_group_conv) for SingleGroup mode, which requires that the output channel per group is a multiple of Threadblock tile N.
+* [Optimized DepthWise Conv](./examples/46_depthwise_simt_conv2dfprop/depthwise_simt_conv2dfprop.cu).  Two new modes are added
+  * [kOptimized](./test/unit/conv/device/depthwise_conv2d_fprop_direct_conv_f16nhwc_f16nhwc_f16nhwc_simt_f16_sm60.cu) - use direct conv to compute instead of implicit GEMM. 
+    *  The restrictions are: 1) input ,output channel and group number should be multiple of (128 / sizeof(input element)). 2) The input filter size should be the same as the template parameter configuration.
+  * [kFixedStrideDilation](./test/unit/conv/device/depthwise_conv2d_fprop_direct_conv_fixed_stride_dilation_f16nhwc_f16nhwc_f16nhwc_simt_f16_sm60.cu) - which puts stride and dilation into templates to further improve the performance. In this mode, kernel persistents some inputs into register to squeeze more performance, so large filter/stride/dilation is not recommanded.
+    * The restrictions are: 1) input, output channel and group number should be multiple of (128 / sizeof(input element)). 2) input filter size, stride, dilation should same as the template parameter configuration. 
+* [Scripts](./examples/44_multi_gemm_ir_and_codegen) to fuse multiple back-to-back GEMM.  Its implementation was discussed in a GTC'22 Spring [talk](https://www.nvidia.com/en-us/on-demand/session/gtcspring22-s41606/).
+* [FP8 data type definition](./include/cutlass/float8.h) and [conversion routines](./include/cutlass/numeric_conversion.h#L1274-2115).
+* Updates and bugfixes from the community (thanks!).  Big shout out to Meta's [xFormers](https://github.com/facebookresearch/xformers).
+
+* **Deprecation announcement:** CUTLASS plans to deprecate the following:
+  * Maxwell and Pascal GPU architectures
+  * Ubuntu 16.04
+  * CUDA 10.2
+
+## [2.10.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.10.0) (2022-08-23)
+* [CUTLASS Python](./examples/40_cutlass_py) now supports GEMM, CONV, Group GEMM for different data types as well as different epilogue flavours.  
+* Optimizations for CUTLASS's [Grouped GEMM](./examples/24_gemm_grouped/gemm_grouped.cu) kernel.  Threadblock scheduling part is improved.  Some computation can be moved to the host side if applicable.  [Grouped Syr2k](./examples/38_syr2k_grouped/syr2k_grouped.cu) kernels are added, too.
+* Optimizations for [GEMM+Softmax](./examples/35_gemm_softmax).  All the reduction computation is fused into the previous GEMM.  More template arguments are provided to fine tune the performance.
+* [Grouped GEMM for Multihead Attention](./examples/41_multi_head_attention).  This general group gemm based MHA does not require the sequence length of all GEMMs to be the same which makes it most useful for natural language processing.
+* [GEMM + Layer norm fusion for Ampere](./examples/37_gemm_layernorm_gemm_fusion/) splits the layernorm into two parts and both of them can be fused into the GEMMs before and after separately.  In addition to use square sum to compute variance of layernorm, [Shift-K](https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Computing_shifted_data) is provided if square sum raise numerical issues.
+* [GEMM Epilogue Permutation Fusion](./examples/39_gemm_permute) can apply user provided permutation layout mapping in the GEMM epilogue.
+* [Grouped convolution targeting implicit GEMM](test/unit/conv/device/group_conv2d_fprop_implicit_gemm_f16nhwc_f16nhwc_f16nhwc_tensor_op_f32_sm80.cu) introduces the first group convolution implementation to CUTLASS.  It is an Analytical implementation, not an Optimized.  The restrictions are: 1) input and output channel number should be multiple of group number. 2) split-K is not supported.  The implementation has 2 modes: 
+  * kSingleGroup: output channel per group is multiple of Threadblock tile N.
+  * kMultipleGroup: Threadblock tile N is multiple of output channel per group.
+* [Depthwise separable convolution](test/unit/conv/device/depthwise_conv2d_fprop_implicit_gemm_f16nhwc_f16nhwc_f16nhwc_simt_f16_sm60.cu) introduces the first depthwise convolution which is also Analytical for now.  The restrictions are: 1) SIMT only 2) No split-K 3) input channel equals to output channel equals to group number.
+* Standalone [Layernorm](./tools/util/include/cutlass/util/device_layernorm.h) and [Pooling](./tools/util/include/cutlass/util/device_nhwc_pooling.h) kernels.
+* [Back-to-back GEMM/CONV](./examples/13_two_tensor_op_fusion) relaxes the requirement that the first GEMM K dimension needs to be the multiple of Threadblock Tile K dimension.
+* Optimal performance using [**CUDA 11.6u2**](https://developer.nvidia.com/cuda-downloads)
+* Updates and bugfixes from the community (thanks!)
+
+## [2.9.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.9.0) (2022-04-21)
+
+* [First layer Convolution kernels](./test/unit/conv/device/conv2d_fprop_fixed_channels_f16nhwc_f16nhwc_f16nhwc_tensor_op_f32_sm80.cu) specialized for small channel counts and reduced alignment
+  * [Few channels](./include/cutlass/conv/threadblock/conv2d_fprop_activation_tile_access_iterator_few_channels.h) specialization for reduced alignment capabilities
+  * [Fixed channels](./include/cutlass/conv/threadblock/conv2d_fprop_activation_tile_access_iterator_fixed_channels.h) further specialized when channel count perfectly matches the access vector size
+  * [Unit tests](./test/unit/conv/device/conv2d_fprop_few_channels_f16nhwc_f16nhwc_f16nhwc_tensor_op_f32_sm80.cu)
+  * [Python-based instance emitter](./python/cutlass_library/generator.py) in the CUTLASS Library and support in the Profiler
+* [BLAS3](https://docs.nvidia.com/cuda/cublas/index.html#cublas-level-3-function-reference) operators accelerated by Tensor Cores
+  * Supported types: f32, cf32, f64, cf64, tf32x3, complex tf32x3
+  * [HERK](./test/unit/gemm/device/her2k_cf32h_cf32n_tensor_op_fast_f32_sm80.cu) with [emitter](./python/cutlass_library/rank_k_operation.py)
+  * [SYRK](./test/unit/gemm/device/syrk_f32n_f32t_tensor_op_fast_f32_sm80.cu) with [emitter](./python/cutlass_library/rank_k_operation.py)
+  * [SYMM](./test/unit/gemm/device/symm_f32n_f32n_tensor_op_fast_f32_ls_sm80.cu) with [emitter](./python/cutlass_library/symm_operation.py)
+  * [TRMM](./test/unit/gemm/device/trmm_f32n_f32t_f32t_tensor_op_fast_f32_ls_sm80.cu) with [emitter](./python/cutlass_library/trmm_operation.py)
+  * [Unit tests](./test/unit/gemm/device/testbed_rank_k_universal.h)
+* [CUTLASS Python](./examples/40_cutlass_py) demonstrating JIT compilation of CUTLASS kernels and a Python-based runtime using [CUDA Python](https://developer.nvidia.com/cuda-python)
+  * [Python-based runtime](./tools/library/scripts/rt.py) interoperable with existing emitters
+* [GEMM + Softmax example](./examples/35_gemm_softmax)
+* [Gather and Scatter Fusion with GEMM](./examples/36_gather_scatter_fusion) can gather inputs and scatters outputs based on indices vectors in the same GEMM kernel.
+  * It can select random rows in a row major matrix.
+  * It can select random columns in a column major matrix.
+* [Back-to-back GEMM/CONV](./examples/13_two_tensor_op_fusion) fully supports buffering the first GEMM/CONV results in the shared memory for the latter one to use.  It can eliminate register spill when the tile size is big.  Additionally, bias vector add is supported in the first GEMM/CONV.
+  * Supported kernels: GEMM and CONV.
+  * Supported types: fp16 and int8.
+  * Supported architectures: Turing and Ampere.
+* [Transposed Convolution](./examples/34_transposed_conv2d) (a.k.a Deconvolution) support which reuses Dgrad implementation.
+* [Utility functions](./tools/util/include/cutlass/util) that can pad NHWC and convert between NCHW and NHWC.
+* [Small alignment implicit gemm](https://github.com/NVIDIA/cutlass/issues/242) support for Fprop/Dgrad/Wgrad so that padding is no longer mandated to use tensor cores in these kernels.
+* Epilogue enhancement:
+  * Eliminate bank conflicts in int8 tensor core kernels.
+  * Half2 usage if epilogue compute type is fp16.
+  * More activation functions: Silu, Hardswish, Leaky Relu.
+  * New elementwise fusion pattern for [residual block](./include/cutlass/epilogue/thread/linear_combination_residual_block.h).
+* [Group GEMM](./examples/24_gemm_grouped) thread block number calculation fix which helps to launch the intended number of threadblocks to fully occupy the GPUs.
+* [Parallel GEMM splitk](https://github.com/NVIDIA/cutlass/pull/277) support in the CUTLASS profiler.
+* Optimal performance using [**CUDA 11.6u2**](https://developer.nvidia.com/cuda-downloads)
+* Updates and bugfixes from the community (thanks!)
+
+
+## [2.8.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.8.0) (2021-11-19)
+
+* **TF32x3:** emulated single-precision using Tensor Cores
+  * 45+ TFLOPs on NVIDIA A100
+  * [GEMM SDK example](./examples/27_ampere_3xtf32_fast_accurate_tensorop_gemm/27_ampere_3xtf32_fast_accurate_tensorop_gemm.cu) (real)
+  * [COMPLEX GEMM SDK example](./examples/29_ampere_3xtf32_fast_accurate_tensorop_complex_gemm/29_3xtf32_complex_gemm.cu) (complex)
+  * [Implicit GEMM Convolution SDK example](./examples/28_ampere_3xtf32_fast_accurate_tensorop_fprop/ampere_3xtf32_fast_accurate_tensorop_fprop.cu)
+* **Mainloop fusion for Convolution:** convolution with fused per-channel scale-bias-relu
+  * [Conv Fprop SDK example](./examples/25_ampere_fprop_mainloop_fusion/ampere_fprop_mainloop_fusion.cu)
+  * [Conv WGrad SDK example](./examples/26_ampere_wgrad_mainloop_fusion/ampere_wgrad_mainloop_fusion.cu) 
+  * [cutlass::conv::device::ImplicitGemmConvolutionFusion](./include/cutlass/conv/device/implicit_gemm_convolution_fusion.h)
+* **Grouped GEMM:** similar to batched GEMM with distinct problem size per group
+  * [SDK example](./examples/24_gemm_grouped) with performance comparison with Batched Strided GEMM
+  * [cutlass::gemm::device::GemmGrouped](./include/cutlass/gemm/device/gemm_grouped.h)
+* [Implicit GEMM Convolution fusion](./examples/13_two_tensor_op_fusion/) supports staging 1st convolution's output accumulator in the shared memory on Turing. This allows more flexible warp tile sizes and less regsiter pressue.
+* Optimal performance using [**CUDA 11.5**](https://developer.nvidia.com/cuda-downloads)
+* Updates from the community (thanks!)
+
+* **Deprecation announcement:** CUTLASS plans to deprecate the following:
+  * Maxwell and Pascal GPU architectures
+  * Ubuntu 16.04
+  * CUDA 10.2
+
+## [2.7.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.7.0) (2021-09-24)
+  * Mainloop fusion for GEMM: [summation over A or B](./examples/23_ampere_gemm_operand_reduction_fusion/ampere_gemm_operand_reduction_fusion.cu)
+  * [Strided DGRAD (optimized iterators)](./include/cutlass/conv/kernel/default_conv2d_dgrad.h)
+  * [Half-precision GELU_taylor activation functions](./include/cutlass/epilogue/thread/activation.h#L196)
+    * Use these when accumulation and epilogue compute types are all `cutlass::half_t`
+  * Tuning and bug fixes to [fused GEMM + GEMM example](./examples/13_two_tensor_op_fusion/)
+  * Support for smaller than 128b aligned Convolutions: [see examples](test/unit/conv/device/conv2d_fprop_implicit_gemm_f16nhwc_f16nhwc_f16nhwc_tensor_op_f16_sm80.cu#L272)
+  * Caching of results to accelerate Convolution [unit tests](test/unit/conv/device/cache_testbed_output.h)
+    * Can be enabled or disabled by running `cmake .. -DCUTLASS_TEST_ENABLE_CACHED_RESULTS=OFF`
+  * Corrections and bug fixes reported by the CUTLASS community
+    * Thank you for filing these issues!
+
+## [2.6.1](https://github.com/NVIDIA/cutlass/releases/tag/v2.6.1) (2021-09-03)
+  * Arbitrary padding and striding for CUTLASS Strided DGRAD Convolution operator (Analytic Iterators)
+  * Tuning for GEMMs fused with partial reductions
+  * Corrections and bug fixes reported by the CUTLASS community
+    * Thank you for filing these issues!
+
+## [2.6.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.6.0) (2021-07-22)
+  * Optimal performance when compiled with the [CUDA 11.4 Toolkit](https://developer.nvidia.com/cuda-toolkit)
+    * Adopt the new L2 prefetch feature in [cp.async](./include/cutlass/arch/memory.h) and [global load](./include/cutlass/arch/memory_sm80.h)
+  * Fused operators with GEMM and Convolution
+    * [Fused broadcast in epilogue](test/unit/gemm/device/gemm_with_broadcast_f16n_f16n_f16n_tensorop_f32_sm75.cu)
+    * [Fused partial reduction in epilogue](./test/unit/gemm/device/gemm_with_reduction_f16n_f16n_f16n_tensorop_f32_sm75.cu)
+  * 64b tensor strides and leading dimensions support for GEMMs
+  * Affine rank=2 matrix layouts 
+    * Row stride and column stride for matrices using [cutlass::layout::AffineRank2](./include/cutlass/layout/matrix.h)
+    * Support [FP64 tensor core](./examples/18_ampere_fp64_tensorop_affine2_gemm/ampere_fp64_tensorop_affine2_gemm.cu) and SIMT GEMM.
+  * [Batched GEMV](./test/unit/gemm/device/gemv.cu) preview implementation
+  * [New strided Dgrad](test/unit/conv/device/conv2d_strided_dgrad_implicit_gemm_f16nhwc_f16nhwc_f32nhwc_tensor_op_f32_sm80.cu) implementation
+    * Accelerates over previous implementation by cutting down redundant math by 4x
+    * Support using new `Dy` and `w` analytic iterators and existing `cutlass::conv::device::ImplicitGemmConvolution` interface
+  * Quaternion-valued GEMM and Convolution in single- and double-precision (targeting CUDA Cores)
+    * Updates to [quaternion.h](./include/cutlass/quaternion.h) and [functional.h](./include/cutlass/functional.h)
+    * SDK Example for [GEMM](./examples/21_quaternion_gemm/quaternion_gemm.cu) and [Convolution](./examples/22_quaternion_conv/quaternion_conv.cu)
+    * [Unit tests for GEMM](./test/unit/gemm/device/simt_qgemm_nn_sm50.cu) and [Convolution](./test/unit/conv/device/conv2d_fprop_implicit_gemm_qf32nhwc_qf32nhwc_qf32nhwc_simt_f32_sm50.cu)
+  * Many improvements to the epilogue.
+    * Provide an [option](./include/cutlass/epilogue/threadblock/epilogue.h) to not fully unroll the epilogue to reduce the code size and improve the performance when using complicated elementwise operations
+    * Performance improvement for FP16 tensor core kernels
+    * Bug fixes
+  * Enhanced Clang support and the combination of Clang 13 and CUDA 11.4 can build and run kernels from Pascal and Ampere.   
+  * Updated minimum CUDA Toolkit requirement to 10.2
+    * [CUDA 11.4 Toolkit](https://developer.nvidia.com/cuda-toolkit) recommended
+  * Corrections and bug fixes reported by the CUTLASS community
+    * Thank you for filing these issues!
+
+## [2.5.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.5.0) (2021-02-26)
+  * Tensor reductions
+    * _m_-to-_n_ reductions of tensors with affine layout
+    * [Specializations](./test/unit/reduction/device/tensor_reduce_contiguous.cu) for reductions including contiguous dimension
+    * [Specializations](./test/unit/reduction/device/tensor_reduce_strided.cu) for reductions excluding contiguous dimension
+    * Custom reduction functors such as `cutlass::logical_and`
+    * Large tensor support, up to 2^63 elements (however, each dimension is limited to an extent of 2^31)
+  * Optimizations for 3-D convolution
+    * [Optimized tile iterators](./include/cutlass/conv/threadblock/conv3d_fprop_activation_tile_access_iterator_optimized.h) using precomputed delta table for 3-D convolution
+    * Full coverage of [forward](test/unit/conv/device/conv3d_fprop_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm80.cu) and [backwards](test/unit/conv/device/conv3d_dgrad_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm80.cu) passes for 3D convolution
+  * [Fused Convolution+Convolution example](./examples/13_two_tensor_op_fusion/README.md)
+  * Corrections and bug fixes reported by the CUTLASS community
+    * Thank you for filing these issues!
+  
+
+## [2.4.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.4.0) (2020-11-19)
+  * Implicit GEMM convolution kernels supporting CUDA and Tensor Cores on NVIDIA GPUs
+    * Operators: forward (Fprop), backward data gradient (Dgrad), and backward weight gradient (Wgrad) convolution
+    * Data type: FP32, complex<FP32>, Tensor Float 32 (TF32), BFloat16 (BF16), Float16, Int4, Int8, Int32
+    * Spatial dimensions: 1-D, 2-D, and 3-D
+    * Layout: NHWC, NCxHWx
+  * Implicit GEMM convolution components: 
+    * Global memory iterators supporting Fprop, Dgrad, and Wgrad
+    * `MmaMultistage` for implicit GEMM convolution for NVIDIA Ampere architecture
+    * `MmaPipeline` for implicit GEMM convolution for NVIDIA Volta and Turing architectures
+    * [Documentation](./media/docs/implicit_gemm_convolution.md) describing Implicit GEMM Convolution algorithm and implementation
+
+## [2.3.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.3.0) (2020-09-23)
+ * [NVIDIA Ampere Architecture features](https://devblogs.nvidia.com/nvidia-ampere-architecture-in-depth/)
+   * [Sparse Tensor Core GEMM kernels](test/unit/gemm/device/gemm_f16n_f16n_f32t_tensor_op_f32_sparse_sm80.cu):
+     * Direct access to Sparse Tensor Cores and maximum performance via [`mma.sp.sync`](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#warp-level-matrix-instructions-mma-and-friends)
+   * Fast SGEMM targeting GeForce RTX 30-series CUDA Cores
+ * Minor Features:
+   * [Activation functions](./include/cutlass/epilogue/thread/activation.h) such as [GeLU](./include/cutlass/epilogue/thread/linear_combination_gelu.h) and [Sigmoid](./include/cutlass/epilogue/thread/linear_combination_sigmoid.h)
+   * Small [matrix](./include/cutlass/matrix.h) and [quaternion](./include/cutlass/quaternion.h) template classes in device code
+   * [Floating-point constants](./include/cutlass/constants.h)
+ * NVIDIA Ampere GPU Architecture examples and documentation:
+   * [Tensor Float 32](./examples/14_ampere_tf32_tensorop_gemm/ampere_tf32_tensorop_gemm.cu) and 
+   * [Sparse Tensor Cores](./examples/15_ampere_sparse_tensorop_gemm/ampere_sparse_tensorop_gemm.cu)
+   * Documentation added on CUTLASS [efficient row-major epilogue](./media/docs/gemm_api.md#efficient-epilogue)
+
+## [2.2.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.2.0) (2020-06-08)
+ * [NVIDIA Ampere Architecture features](https://devblogs.nvidia.com/nvidia-ampere-architecture-in-depth/)
+   * Fast Tensor Core operations: 
+    * Maximum performance via [`mma.sync`](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#warp-level-matrix-instructions-mma-and-friends)
+    * Tensor Float 32, BFloat16, and double-precision data types
+    * Mixed integer data types (int8, int4, bin1)
+   * Asynchronous copy for deep software pipelines via [`cp.async`](https://docs.nvidia.com/cuda/parallel-thread-execution)   
+   * Described in [GTC 2020 Webinar (SR 21745)](https://developer.nvidia.com/gtc/2020/video/s21745) (free registration required)
+ * Features:
+   * SDK examples showing GEMM fused with bias+relu and fused GEMM+GEMM
+   * Complex-valued GEMMs targeting NVIDIA Ampere Tensor Cores in double-precision and Tensor Float 32
+   * Gaussian complex GEMMs using 3m complex multiply algorithm
+   * Universal GEMM kernel supporting two batch modes and two algorithms for parallel reductions
+ * Policy updates:
+   * [CUDA 11 Toolkit](https://developer.nvidia.com/cuda-toolkit) needed to enable NVIDIA Ampere Architecture features
+   * Disabled F16C by default for compatibility - enable on cmake command line with `-DCUTLASS_ENABLE_F16C=ON`
+
+## [2.1.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.1.0) (2020-04-06)
+ * BLAS-style host-side API added to [CUTLASS Library](./media/docs/quickstart.md#cutlass-library)
+    * API to launch compiled kernel instances for GEMM and planar complex GEMM
+ * Planar Complex GEMM kernels targeting Volta and Turing Tensor Cores
+    * Computes complex matrix products on matrices stored as disjoint real and imaginary parts
+    * [SDK Examples of Planar Complex GEMMs](./examples/10_planar_complex/planar_complex.cu)
+ * Minor enhancements and bug fixes
+
+## [2.0.0](https://github.com/NVIDIA/cutlass/releases/tag/v2.0.0) (2019-11-19)
+ * Substantially refactored for
+    * Better performance, particularly for native Turing Tensor Cores
+    * Robust and durable templates spanning the design space
+    * Encapsulated functionality embodying modern C++11 programming techniques
+    * Optimized containers and data types for efficient, generic, portable device code
+  * Updates to:
+    * [Quick start guide](./media/docs/quickstart.md)
+    * [Documentation](./README.md#documentation)
+    * [Utilities](./media/docs/utilities.md)
+    * [CUTLASS Profiler](./media/docs/profiler.md)
+ * Native Turing Tensor Cores
+    * Efficient GEMM kernels targeting Turing Tensor Cores
+    * Mixed-precision floating point, 8-bit integer, 4-bit integer, and binarized operands
+ * Coverage of existing CUTLASS functionality
+    * GEMM kernels targeting CUDA and Tensor Cores in NVIDIA GPUs
+    * Volta Tensor Cores through native mma.sync and through WMMA API
+    * Optimizations such as parallel reductions, threadblock rasterization, and intra-threadblock reductions
+    * Batched GEMM operations
+    * Complex-valued GEMMs
+ * **Note: a host compiler supporting C++11 or greater is required.**
+
+# CUTLASS 1.x
+
+## [1.3.2](https://github.com/NVIDIA/cutlass/releases/tag/v1.3.2) (2019-07-09)
+ * Performance improvement for Volta Tensor Cores TN and TT layouts.
+
+## [1.3.1](https://github.com/NVIDIA/cutlass/releases/tag/v1.3.1) (2019-04-09)
+ * Corrected NVRTC unit tests.
+
+## [1.3.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.3.0) (2019-03-20)
+ * Efficient GEMM kernel targeting Volta Tensor Cores via `mma.sync` instruction added in CUDA 10.1.
+
+## [1.2.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.2.0) (2018-10-26)
+ * Parallelized reductions across threadblocks ("Split-K")
+   * Improved IGEMM performance
+ * Batched strided WMMA GEMMs
+
+## [1.1.0](https://github.com/NVIDIA/cutlass/releases/tag/v1.1.0) (2018-09-19)
+  * Turing Features
+    * WMMA GEMM targeting TensorCores - INT8, INT4, 1-bit
+  * Batched Strided GEMM
+  * Threadblock rasterization strategies
+    * Improved performance for adverse problem sizes and data layouts
+  * Extended CUTLASS Core comonents
+    * Tensor views support arbitrary matrix and tensor layouts
+    * Zip iterators for structuring multiple data streams
+  * Enhanced CUTLASS utilities
+    * Reference code for tensor operations in host and device code
+    * Added HostMatrix<> for simplified matrix creation
+  * Examples
+    * Basic GEMM, tensor views, CUTLASS utilities, batched GEMM, WMMA GEMM
+
+## [1.0.1](https://github.com/NVIDIA/cutlass/releases/tag/v1.0.1) (2018-06-11)
+
+  * Intra-threadblock reduction added for small threadblock tile sizes
+    * sgemm_64x128x16, sgemm_128x128x16, sgemm_128x64x16, sgemm_128x32x16, sgemm_64x64x16, sgemm_64x32x16
+    * igemm_32x32x128
+  * GEMM _K_ residue handled during prologue prior to mainloop
+  * Replaced Google Test copy with submodule. Use `git submodule init --recursive --update`
+
+## [1.0.0](https://github.com/NVIDIA/cutlass/commit/2028ebe120aab22bfd0b2baf8902d4c9627eb33f) (2018-05-16)
+
+  * Substantial rewrite to accommodate new architecture
+  * Kernels: SGEMM, DGEMM, IGEMM, HGEMM, WMMA GEMM
+  * Unit and performance tests
+
+## [0.0.1](https://github.com/NVIDIA/cutlass/commit/d08ba8ac46e2fa3f745e070c390182edb56b2e91) (2017-12-04)
+
+  * Initial release
+
+
+## Copyright
+
+Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+SPDX-License-Identifier: BSD-3-Clause
+
+```
+  Redistribution and use in source and binary forms, with or without
+  modification, are permitted provided that the following conditions are met:
+
+  1. Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+  2. Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+  3. Neither the name of the copyright holder nor the names of its
+  contributors may be used to endorse or promote products derived from
+  this software without specific prior written permission.
+
+  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+  AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+  IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+  DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+  DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+  SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+  CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+  OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+  OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+```
+

crates/vendor/candle/candle-flash-attn/cutlass/CITATION.cff

@@ -0,0 +1,112 @@
+cff-version: 1.2.0
+title: CUTLASS
+message: >-
+  If you use this software, please cite using the
+  following metadata.
+type: software
+authors:
+  - given-names: Vijay
+    family-names: Thakkar
+    email: vithakkar@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Pradeep
+    family-names: Ramani
+    email: prramani@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Cris
+    family-names: Cecka
+    email: ccecka@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Aniket
+    family-names: Shivam
+    email: ashivam@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Honghao
+    family-names: Lu
+    email: honghaol@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Ethan
+    family-names: Yan
+    email: etyan@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Jack
+    family-names: Kosaian
+    email: jkosaian@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Mark
+    family-names: Hoemmen
+    email: mhoemmen@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Haicheng
+    family-names: Wu
+    email: haichengw@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Andrew
+    family-names: Kerr
+    email: akerr@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Matt
+    family-names: Nicely
+    email: mnicely@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Duane
+    family-names: Merrill
+    email: dumerrill@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Dustyn
+    family-names: Blasig
+    email: dblasig@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Fengqi
+    family-names: Qiao
+    email: fqiao@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Piotr
+    family-names: Majcher
+    email: pmajcher@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Paul
+    family-names: Springer
+    email: pspringer@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Markus
+    family-names: Hohnerbach
+    affiliation: NVIDIA
+    email: mhohnerbach@nvidia.com
+  - given-names: Jin
+    family-names: Wang
+    email: jinw@nvidia.com
+    affiliation: NVIDIA
+  - given-names: Manish
+    family-names: Gupta
+    affiliation: Google
+    email: manigupta@google.com
+
+
+repository-code: 'https://github.com/NVIDIA/cutlass'
+abstract: >-
+  CUTLASS is a collection of CUDA C++ template
+  abstractions for implementing high-performance
+  matrix-multiplication (GEMM) and related
+  computations at all levels and scales within CUDA.
+  It incorporates strategies for hierarchical
+  decomposition and data movement similar to those
+  used to implement cuBLAS and cuDNN. CUTLASS
+  decomposes these "moving parts" into reusable,
+  modular software components abstracted by C++
+  template classes. These thread-wide, warp-wide,
+  block-wide, and device-wide primitives can be
+  specialized and tuned via custom tiling sizes, data
+  types, and other algorithmic policy. The resulting
+  flexibility simplifies their use as building blocks
+  within custom kernels and applications.
+keywords:
+  - 'cutlass, tensor cores, cuda, cute, nvidia, gpu, linear algebra, matrix computations'
+license: BSD-3-Clause
+license-url: https://github.com/NVIDIA/cutlass/blob/v3.0.0/LICENSE.txt
+version: '3.0.0'
+date-released: '2023-01-23'
+identifiers:
+  - type: url
+    value: "https://github.com/NVIDIA/cutlass/tree/v3.0.0"
+    description: The GitHub release URL of tag 3.0.0

crates/vendor/candle/candle-flash-attn/cutlass/CONTRIBUTORS.md

@@ -0,0 +1,87 @@
+![ALT](./media/images/gemm-hierarchy-with-epilogue-no-labels.png "CUTLASS")
+
+[README](./README.md#documentation) > **Contributors**
+
+# CUTLASS Developers and Contributors
+
+This is the official list of CUTLASS developers and contributors.
+
+## DEVELOPERS
+Vijay Thakkar<br />
+Pradeep Ramani<br />
+Cris Cecka<br />
+Aniket Shivam<br />
+Jack Kosaian<br />
+Mark Hoemmen<br />
+Richard Cai<br />
+Honghao Lu<br />
+Ethan Yan<br />
+Haicheng Wu<br />
+Andrew Kerr<br />
+Dustyn Blasig<br />
+Fengqi Qiao<br />
+Duane Merrill<br />
+Yujia Zhai<br />
+Rawn Henry<br />
+Sergey Klevtsov<br />
+Shang Zhang<br />
+Piotr Majcher<br />
+Paul Springer<br />
+Markus Hohnerbach<br />
+Jin Wang<br />
+Aditya Atluri<br />
+
+## CuTe
+Cris Cecka<br />
+Vijay Thakkar<br />
+
+## CUTLASS Product Manager
+Matthew Nicely<br />
+
+## Former CUTLASS Developers
+Manish Gupta<br />
+Naila Farooqui<br />
+David Tanner<br />
+Manikandan Ananth<br />
+Zhaodong Chen<br />
+Chinmay Talegaonkar<br />
+
+## CONTRIBUTORS
+Timothy Costa<br />
+Julien Demouth<br />
+Brian Fahs<br />
+Michael Garland<br />
+Michael Goldfarb<br />
+Mostafa Hagog<br />
+Fei Hu<br />
+Alan Kaatz<br />
+Tina Li<br />
+Timmy Liu<br />
+Wei Liu<br />
+Tim Martin<br />
+Duane Merrill<br />
+Kevin Siu<br />
+Markus Tavenrath<br />
+John Tran<br />
+Vicki Wang<br />
+Junkai Wu<br />
+Fung Xie<br />
+Albert Xu<br />
+Yang Xu<br />
+Jack Yang<br />
+Scott Yokim<br />
+Xiuxia Zhang<br />
+Nick Zhao<br />
+
+## ACKNOWLEDGEMENTS
+
+Girish Bharambe<br />
+Luke Durant<br />
+Carter Edwards<br />
+Olivier Giroux<br />
+Stephen Jones<br />
+Rishkul Kulkarni<br />
+Bryce Lelbach<br />
+Joel McCormack<br />
+Kyrylo Perelygin<br />
+Sean Treichler<br />

crates/vendor/candle/candle-flash-attn/cutlass/CUDA.cmake

@@ -0,0 +1,394 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+if(CUDA_COMPILER MATCHES "[Cc]lang")
+  set(CUTLASS_NATIVE_CUDA_INIT ON)
+elseif(CMAKE_VERSION VERSION_LESS 3.12.4)
+  set(CUTLASS_NATIVE_CUDA_INIT OFF)
+else()
+  set(CUTLASS_NATIVE_CUDA_INIT ON)
+endif()
+
+set(CUTLASS_NATIVE_CUDA ${CUTLASS_NATIVE_CUDA_INIT} CACHE BOOL "Utilize the CMake native CUDA flow")
+
+if(NOT DEFINED ENV{CUDACXX} AND NOT DEFINED ENV{CUDA_BIN_PATH} AND DEFINED ENV{CUDA_PATH})
+  # For backward compatibility, allow use of CUDA_PATH.
+  set(ENV{CUDACXX} $ENV{CUDA_PATH}/bin/nvcc)
+endif()
+
+if(CUTLASS_NATIVE_CUDA)
+
+  enable_language(CUDA)
+
+  if(NOT CUDA_VERSION)
+    set(CUDA_VERSION ${CMAKE_CUDA_COMPILER_VERSION})
+  endif()
+  if(NOT CUDA_TOOLKIT_ROOT_DIR)
+    get_filename_component(CUDA_TOOLKIT_ROOT_DIR "${CMAKE_CUDA_COMPILER}/../.." ABSOLUTE)
+  endif()
+
+else()
+
+  find_package(CUDA REQUIRED)
+  # We workaround missing variables with the native flow by also finding the CUDA toolkit the old way.
+
+  if(NOT CMAKE_CUDA_COMPILER_VERSION)
+    set(CMAKE_CUDA_COMPILER_VERSION ${CUDA_VERSION})
+  endif()
+
+endif()
+
+if (CUDA_VERSION VERSION_LESS 9.2)
+  message(FATAL_ERROR "CUDA 9.2+ Required, Found ${CUDA_VERSION}.")
+endif()
+if(NOT CUTLASS_NATIVE_CUDA OR CUDA_COMPILER MATCHES "[Cc]lang")
+  set(CMAKE_CUDA_COMPILER ${CUDA_TOOLKIT_ROOT_DIR}/bin/nvcc)
+  message(STATUS "CUDA Compiler: ${CMAKE_CUDA_COMPILER}")
+endif()
+
+find_library(
+  CUDART_LIBRARY cudart
+  PATHS
+  ${CUDA_TOOLKIT_ROOT_DIR}
+  PATH_SUFFIXES
+  lib/x86_64-linux-gnu
+  lib/x64
+  lib64
+  lib
+  NO_DEFAULT_PATH
+  # We aren't going to search any system paths. We want to find the runtime 
+  # in the CUDA toolkit we're building against.
+  )
+
+if(NOT TARGET cudart AND CUDART_LIBRARY)
+
+  message(STATUS "CUDART: ${CUDART_LIBRARY}")
+
+  if(WIN32)
+    add_library(cudart STATIC IMPORTED GLOBAL)
+    # Even though we're linking against a .dll, in Windows you statically link against
+    # the .lib file found under lib/x64. The .dll will be loaded at runtime automatically
+    # from the PATH search.
+  else()
+    add_library(cudart SHARED IMPORTED GLOBAL)
+  endif()  
+
+  add_library(nvidia::cudart ALIAS cudart)
+  
+  set_property(
+    TARGET cudart
+    PROPERTY IMPORTED_LOCATION
+    ${CUDART_LIBRARY}
+    )
+
+elseif(TARGET cudart)
+
+  message(STATUS "CUDART: Already Found")
+
+else()
+
+  message(STATUS "CUDART: Not Found")
+
+endif()
+
+find_library(
+  CUDA_DRIVER_LIBRARY cuda
+  PATHS
+  ${CUDA_TOOLKIT_ROOT_DIR}
+  PATH_SUFFIXES
+  lib/x86_64-linux-gnu
+  lib/x64
+  lib64
+  lib
+  lib64/stubs
+  lib/stubs
+  NO_DEFAULT_PATH
+  # We aren't going to search any system paths. We want to find the runtime 
+  # in the CUDA toolkit we're building against.
+  )
+
+if(NOT TARGET cuda_driver AND CUDA_DRIVER_LIBRARY)
+
+  message(STATUS "CUDA Driver: ${CUDA_DRIVER_LIBRARY}")
+
+  if(WIN32)
+    add_library(cuda_driver STATIC IMPORTED GLOBAL)
+    # Even though we're linking against a .dll, in Windows you statically link against
+    # the .lib file found under lib/x64. The .dll will be loaded at runtime automatically
+    # from the PATH search.
+  else()
+    add_library(cuda_driver SHARED IMPORTED GLOBAL)
+  endif()  
+
+  add_library(nvidia::cuda_driver ALIAS cuda_driver)
+  
+  set_property(
+    TARGET cuda_driver
+    PROPERTY IMPORTED_LOCATION
+    ${CUDA_DRIVER_LIBRARY}
+    )
+
+elseif(TARGET cuda_driver)
+
+  message(STATUS "CUDA Driver: Already Found")
+
+else()
+
+  message(STATUS "CUDA Driver: Not Found")
+
+endif()
+
+find_library(
+  NVRTC_LIBRARY nvrtc
+  PATHS
+  ${CUDA_TOOLKIT_ROOT_DIR}
+  PATH_SUFFIXES
+  lib/x64
+  lib64
+  lib
+  NO_DEFAULT_PATH
+  # We aren't going to search any system paths. We want to find the runtime 
+  # in the CUDA toolkit we're building against.
+  )
+
+if(NOT TARGET nvrtc AND NVRTC_LIBRARY)
+
+  message(STATUS "NVRTC: ${NVRTC_LIBRARY}")
+
+  if(WIN32)
+    add_library(nvrtc STATIC IMPORTED GLOBAL)
+    # Even though we're linking against a .dll, in Windows you statically link against
+    # the .lib file found under lib/x64. The .dll will be loaded at runtime automatically
+    # from the PATH search.
+  else()
+    add_library(nvrtc SHARED IMPORTED GLOBAL)
+  endif()  
+  
+  add_library(nvidia::nvrtc ALIAS nvrtc)
+  
+  set_property(
+    TARGET nvrtc
+    PROPERTY IMPORTED_LOCATION
+    ${NVRTC_LIBRARY}
+    )
+
+elseif(TARGET nvrtc)
+
+  message(STATUS "NVRTC: Already Found")
+
+else()
+
+  message(STATUS "NVRTC: Not Found")
+
+endif()
+
+include_directories(SYSTEM ${CUDA_INCLUDE_DIRS})
+# Some platforms (e.g. Visual Studio) don't add the CUDA include directories to the system include
+# paths by default, so we add it explicitly here.
+
+function(cutlass_correct_source_file_language_property)
+  if(CUDA_COMPILER MATCHES "[Cc]lang")
+    foreach(File ${ARGN})
+      if(File MATCHES ".*\.cu$")
+        set_source_files_properties(${File} PROPERTIES LANGUAGE CXX)
+      endif()
+    endforeach()
+  endif()
+endfunction()
+
+if (MSVC OR CUTLASS_LIBRARY_KERNELS MATCHES "all")
+  set(CUTLASS_UNITY_BUILD_ENABLED_INIT ON)
+else()
+  set(CUTLASS_UNITY_BUILD_ENABLED_INIT OFF)
+endif()
+
+set(CUTLASS_UNITY_BUILD_ENABLED ${CUTLASS_UNITY_BUILD_ENABLED_INIT} CACHE BOOL "Enable combined source compilation")
+
+if (MSVC)
+  set(CUTLASS_UNITY_BUILD_BATCH_SIZE_INIT 8)
+else()
+  set(CUTLASS_UNITY_BUILD_BATCH_SIZE_INIT 16)
+endif()
+
+set(CUTLASS_UNITY_BUILD_BATCH_SIZE ${CUTLASS_UNITY_BUILD_BATCH_SIZE_INIT} CACHE STRING "Batch size for unified source files")
+
+function(cutlass_unify_source_files TARGET_ARGS_VAR)
+
+  set(options)
+  set(oneValueArgs BATCH_SOURCES BATCH_SIZE)
+  set(multiValueArgs)
+  cmake_parse_arguments(_ "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
+
+  if (NOT DEFINED TARGET_ARGS_VAR)
+    message(FATAL_ERROR "TARGET_ARGS_VAR parameter is required")
+  endif()
+
+  if (NOT DEFINED __BATCH_SOURCES)
+    set(__BATCH_SOURCES ON)
+  endif()
+
+  if (__BATCH_SOURCES AND NOT DEFINED __BATCH_SIZE)
+    set(__BATCH_SIZE ${CUTLASS_UNITY_BUILD_BATCH_SIZE})
+  endif()
+
+  if (CUTLASS_UNITY_BUILD_ENABLED AND __BATCH_SOURCES AND __BATCH_SIZE GREATER 1)
+
+    set(CUDA_FILE_ARGS)
+    set(TARGET_SOURCE_ARGS)
+    
+    foreach(ARG ${__UNPARSED_ARGUMENTS})
+      if(${ARG} MATCHES ".*\.cu$")
+        list(APPEND CUDA_FILE_ARGS ${ARG})
+      else()
+        list(APPEND TARGET_SOURCE_ARGS ${ARG})
+      endif()
+    endforeach()
+    
+    list(LENGTH CUDA_FILE_ARGS NUM_CUDA_FILE_ARGS)
+    while(NUM_CUDA_FILE_ARGS GREATER 0)
+      list(SUBLIST CUDA_FILE_ARGS 0 ${__BATCH_SIZE} CUDA_FILE_BATCH)
+      string(SHA256 CUDA_FILE_BATCH_HASH "${CUDA_FILE_BATCH}")
+      string(SUBSTRING ${CUDA_FILE_BATCH_HASH} 0 12 CUDA_FILE_BATCH_HASH)
+      set(BATCH_FILE ${CMAKE_CURRENT_BINARY_DIR}/${NAME}.unity.${CUDA_FILE_BATCH_HASH}.cu)
+      message(STATUS "Generating ${BATCH_FILE}")
+      file(WRITE ${BATCH_FILE} "// Unity File - Auto Generated!\n")
+      foreach(CUDA_FILE ${CUDA_FILE_BATCH})
+        get_filename_component(CUDA_FILE_ABS_PATH ${CUDA_FILE} ABSOLUTE)
+        file(APPEND ${BATCH_FILE} "#include \"${CUDA_FILE_ABS_PATH}\"\n")
+      endforeach()
+      list(APPEND TARGET_SOURCE_ARGS ${BATCH_FILE})
+      if (NUM_CUDA_FILE_ARGS LESS_EQUAL __BATCH_SIZE)
+        break()
+      endif()
+      list(SUBLIST CUDA_FILE_ARGS ${__BATCH_SIZE} -1 CUDA_FILE_ARGS)
+      list(LENGTH CUDA_FILE_ARGS NUM_CUDA_FILE_ARGS)
+    endwhile()
+
+  else()
+
+    set(TARGET_SOURCE_ARGS ${__UNPARSED_ARGUMENTS})
+
+  endif()
+
+  set(${TARGET_ARGS_VAR} ${TARGET_SOURCE_ARGS} PARENT_SCOPE)
+
+endfunction()
+function(cutlass_add_library NAME)
+
+  set(options SKIP_GENCODE_FLAGS)
+  set(oneValueArgs EXPORT_NAME)
+  set(multiValueArgs)
+  cmake_parse_arguments(_ "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
+
+  cutlass_unify_source_files(TARGET_SOURCE_ARGS ${__UNPARSED_ARGUMENTS})
+
+  if(CUTLASS_NATIVE_CUDA OR CUDA_COMPILER MATCHES "clang")
+    cutlass_correct_source_file_language_property(${TARGET_SOURCE_ARGS})
+    add_library(${NAME} ${TARGET_SOURCE_ARGS} "")
+  else()
+    set(CUDA_LINK_LIBRARIES_KEYWORD PRIVATE)
+    cuda_add_library(${NAME} ${TARGET_SOURCE_ARGS} "")
+  endif()
+
+  cutlass_apply_standard_compile_options(${NAME})
+  if (NOT __SKIP_GENCODE_FLAGS)
+  cutlass_apply_cuda_gencode_flags(${NAME})
+  endif()
+
+  target_compile_features(
+   ${NAME}
+   INTERFACE
+   cxx_std_11
+   )
+
+  get_target_property(TARGET_TYPE ${NAME} TYPE)
+
+  if (TARGET_TYPE MATCHES "SHARED")
+    set_target_properties(${NAME} PROPERTIES CUDA_RUNTIME_LIBRARY Shared)
+  elseif(TARGET_TYPE MATCHES "STATIC")
+    set_target_properties(${NAME} PROPERTIES CUDA_RUNTIME_LIBRARY Static)
+  endif()
+
+  if(__EXPORT_NAME)
+    add_library(nvidia::cutlass::${__EXPORT_NAME} ALIAS ${NAME})
+    set_target_properties(${NAME} PROPERTIES EXPORT_NAME ${__EXPORT_NAME})
+  endif()
+
+endfunction()
+
+function(cutlass_add_executable NAME)
+
+  set(options)
+  set(oneValueArgs CUDA_RUNTIME_LIBRARY)
+  set(multiValueArgs)
+  cmake_parse_arguments(_ "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
+
+  if (NOT DEFINED __CUDA_RUNTIME_LIBRARY)
+    set(__CUDA_RUNTIME_LIBRARY Shared)
+  endif()
+
+  set(__CUDA_RUNTIME_LIBRARY_ALLOWED None Shared Static)
+  if (NOT __CUDA_RUNTIME_LIBRARY IN_LIST __CUDA_RUNTIME_LIBRARY_ALLOWED)
+    message(FATAL_ERROR "CUDA_RUNTIME_LIBRARY value '${__CUDA_RUNTIME_LIBRARY}' is not in allowed list of '${__CUDA_RUNTIME_LIBRARY_ALLOWED}'")
+  endif()
+
+  cutlass_unify_source_files(TARGET_SOURCE_ARGS ${__UNPARSED_ARGUMENTS})
+
+  if(CUTLASS_NATIVE_CUDA OR CUDA_COMPILER MATCHES "clang")
+    cutlass_correct_source_file_language_property(${TARGET_SOURCE_ARGS})
+    add_executable(${NAME} ${TARGET_SOURCE_ARGS})
+  else()
+    set(CUDA_LINK_LIBRARIES_KEYWORD PRIVATE)
+    cuda_add_executable(${NAME} ${TARGET_SOURCE_ARGS})
+  endif()
+
+  cutlass_apply_standard_compile_options(${NAME})
+  cutlass_apply_cuda_gencode_flags(${NAME})
+
+  target_compile_features(
+   ${NAME}
+   INTERFACE
+   cxx_std_11
+   )
+
+  set_target_properties(${NAME} PROPERTIES CUDA_RUNTIME_LIBRARY ${__CUDA_RUNTIME_LIBRARY})
+
+endfunction()
+
+function(cutlass_target_sources NAME)
+
+  set(options)
+  set(oneValueArgs)
+  set(multiValueArgs)
+  cmake_parse_arguments(_ "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
+
+  cutlass_unify_source_files(TARGET_SOURCE_ARGS ${__UNPARSED_ARGUMENTS})
+  cutlass_correct_source_file_language_property(${TARGET_SOURCE_ARGS})
+  target_sources(${NAME} ${TARGET_SOURCE_ARGS})
+
+endfunction()

crates/vendor/candle/candle-flash-attn/cutlass/LICENSE.txt

@@ -0,0 +1,27 @@
+Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+SPDX-License-Identifier: BSD-3-Clause
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this
+list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice,
+this list of conditions and the following disclaimer in the documentation
+and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+contributors may be used to endorse or promote products derived from
+this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

crates/vendor/candle/candle-flash-attn/cutlass/PUBLICATIONS.md

@@ -0,0 +1,62 @@
+# Publications Using Cutlass
+
+## 2024
+
+- ["ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference"](https://arxiv.org/abs/2410.21465). Hanshi Sun, Li-Wen Chang, Wenlei Bao, Size Zheng, Ningxin Zheng, Xin Liu, Harry Dong, Yuejie Chi, Beidi Chen. _arXiv_, October 2024.
+
+- ["FLUX: Fast Software-based Communication Overlap On GPUs Through Kernel Fusion"](https://arxiv.org/abs/2406.06858). Li-Wen Chang, Wenlei Bao, Qi Hou, Chengquan Jiang, Ningxin Zheng, Yinmin Zhong, Xuanrun Zhang, Zuquan Song, Chengji Yao, Ziheng Jiang, Haibin Lin, Xin Jin, Xin Liu. _arXiv_, June 2024.
+
+- ["EVT: Accelerating Deep Learning Training with Epilogue Visitor Tree"](https://dl.acm.org/doi/10.1145/3620666.3651369). Zhaodong Chen, Andrew Kerr, Richard Cai, Jack Kosaian, Haicheng Wu, Yufei Ding, and Yuan Xie. _Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems_, April 2024.
+
+- ["Faster Neighborhood Attention: Reducing the O(n^2) Cost of Self Attention at the Threadblock Level"](https://arxiv.org/abs/2403.04690). Ali Hassani, Wen-Mei Hwu, Humphrey Shi. _arXiv_, March 2024.
+
+## 2023
+
+- ["A Case Study in CUDA Kernel Fusion: Implementing FlashAttention-2 on NVIDIA Hopper Architecture using the CUTLASS Library"](https://arxiv.org/abs/2312.11918). Ganesh Bikshandi, Jay Shah. _arXiv_, December 2023.
+
+- ["Benchmarking GPU Tensor Cores on General Matrix Multiplication Kernels through CUTLASS"](https://www.mdpi.com/2076-3417/13/24/13022). Xuanteng Huang, Xianwei Zhang, Panfei Yang, Nong Xiao. _Journal of Applied Sciences_, December 2023.
+
+- ["A Speed Odyssey for Deployable Quantization of LLMs"](https://arxiv.org/abs/2311.09550). Qingyuan Li, Ran Meng, Yiduo Li, Bo Zhang, Liang Li, Yifan Lu, Xiangxiang Chu, Yerui Sun, Yuchen Xie. _arXiv_, November 2023.
+
+- ["FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning"](https://arxiv.org/abs/2307.08691). Tri Dao. _Technical Report_, July 2023.
+
+- ["MegaBlocks: Efficient Sparse Training with Mixture-of-Experts"](https://arxiv.org/abs/2211.15841). Trevor Gale, Deepak Narayanan, Cliff Young, Matei Zaharia. _Proceedings of the Sixth Machine Learning and Systems_, May 2023.
+
+- ["ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs"](https://arxiv.org/abs/2210.03052). Yujia Zhai, Chengquan Jiang, Leyuan Wang, Xiaoying Jia, Shang Zhang, Zizhong Chen, Xin Liu, Yibo Zhu. _Proceedings of the 37th IEEE International Parallel & Distributed Processing Symposium (Best Paper)_, May 2023.
+
+- ["A Framework for Fine-Grained Synchronization of Dependent GPU Kernels"](https://arxiv.org/abs/2305.13450). Abhinav Jangda, Saeed Maleki, Maryam Mehri Dehnavi, Madan Musuvathi, Olli Saarikivi. _Computing Research Repository_, May 2023.
+
+- ["Graphene: An IR for Optimized Tensor Computations on GPUs"](https://dl.acm.org/doi/pdf/10.1145/3582016.3582018). Hagedorn, Bastian, Bin Fan, Hanfeng Chen, Cris Cecka, Michael Garland, Vinod Grover. _Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems_, March 2023.
+
+- ["Mixed Precision Post Training Quantization of Neural Networks with Sensitivity Guided Search"](https://arxiv.org/abs/2302.01382). Clemens JS Schaefer, Elfie Guo, Caitlin Stanton, Xiaofan Zhang, Tom Jablin, Navid Lambert-Shirzad, Jian Li, Chiachen Chou, Siddharth Joshi, Yu Emma Wang. _arXiv_, Feburary 2023.
+
+- ["Dynamic N:M Fine-Grained Structured Sparse Attention Mechanism"](https://dl.acm.org/doi/abs/10.1145/3572848.3577500). Zhaodong Chen, Zheng Qu, Yuying Quan, Liu Liu, Yufei Ding, Yuan Xie. _Proceedings of the 28th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming_, Feburary 2023.
+
+- ["Stream-K: Work-centric Parallel Decomposition for Dense Matrix-Matrix Multiplication on the GPU"](https://arxiv.org/abs/2301.03598). Muhammad Osama, Duane Merrill, Cris Cecka, Michael Garland, John D. Owens. _arXiv_, January 2023.
+
+## 2022
+
+- ["GPU Load Balancing"](https://arxiv.org/abs/2212.08964). Muhammad Osama. _Doctoral dissertation, University of California, Davis_, December 2022.
+
+- ["Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production"](https://arxiv.org/abs/2211.10017). Young Jin Kim, Rawn Henry, Raffy Fahim, Hany Hassan Awadalla. _Proceedings of the Third Workshop on Simple and Efficient Natural Language Processing_, December 2022.
+
+- ["Bolt: Bridging the Gap between Auto-tuners and Hardware-native Performance"](https://arxiv.org/abs/2110.15238). Jiarong Xing, Leyuan Wang, Shang Zhang, Jack Chen, Ang Chen, Yibo Zhu. _Proceedings of the 5th MLSys Conference_, August 2022.
+
+- ["Recovering single precision accuracy from Tensor Cores while surpassing the FP32 theoretical peak performance"](https://arxiv.org/abs/2203.03341). Hiroyuki Ootomo, Rio Yokota. _International Journal of High Performance Computing_, March 2022.
+
+- ["Breaking the Computation and Communication Abstraction Barrier in Distributed Machine Learning Workloads"](https://arxiv.org/abs/2105.05720). Abhinav Jangda, Jun Huang, Guodong Liu, Amir Hossein Nodehi Sabet, Saeed Maleki, Youshan Miao, Madanlal Musuvathi, Todd Mytkowicz, Olli Sarikivi. _Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems_, February 2022.
+
+## 2021
+
+- ["Arithmetic-intensity-guided fault tolerance for neural network inference on GPUs"](https://dl.acm.org/doi/abs/10.1145/3458817.3476184). Jack Kosaian, K. V. Rashmi. _Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis_, November 2021.
+
+- ["Real-time Neural Radiance Caching for Path Tracing"](https://dl.acm.org/doi/abs/10.1145/3450626.3459812). Thomas Muller, Fabrice Rousselle, Jan Novak, Alex Keller. _ACM Trans. Graph._, August 2021.
+
+## 2020
+
+- ["Scalable Knowledge Graph Analytics at 136 Petaflop/s"](https://www.computer.org/csdl/proceedings-article/sc/2020/999800a061/1oeORDgCM0g). Ramakrishnan Kannan, Piyush Sao, Hao Lu, Drahomira Herrmannova, Vijay Thakkar,  Robert Patton, Richard Vuduc, Thomas Potok. _Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis_, November 2020.
+
+- ["Accelerating Sparse DNN Models without Hardware-Support via Tile-Wise Sparsity
+"](https://arxiv.org/abs/2008.13006). Cong Guo, Bo Yang Hsueh, Jingwen Leng, Yuxian Qiu, Yue Guan, Zehuan Wang, Xiaoying Jia, Xipeng Li, Minyi Guo, Yuhao Zhu. _Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis_, November 2020.
+
+- ["Strassen's Algorithm Reloaded on GPUs"](https://dl.acm.org/doi/10.1145/3372419). Jianyu Huang, Chenhan D. Yu, Robert A. van de Geijn. _ACM Transactions on Mathematical Software_, March 2020.

crates/vendor/candle/candle-flash-attn/cutlass/README.md

@@ -0,0 +1,570 @@
+![ALT](./media/images/gemm-hierarchy-with-epilogue-no-labels.png "Complete CUDA GEMM decomposition")
+
+# CUTLASS 3.6.0
+
+_CUTLASS 3.6.0 - October 2024_
+
+CUTLASS is a collection of CUDA C++ template abstractions for implementing
+high-performance matrix-matrix multiplication (GEMM) and related computations at all levels 
+and scales within CUDA. It incorporates strategies for hierarchical decomposition and 
+data movement similar to those used to implement cuBLAS and cuDNN.  CUTLASS decomposes 
+these "moving parts" into reusable, modular software components abstracted by C++ template 
+classes.  Primitives for different levels of a conceptual parallelization hierarchy
+can be specialized and tuned via custom tiling sizes, data types,
+and other algorithmic policy. The resulting flexibility simplifies their use
+as building blocks within custom kernels and applications.
+
+To support a wide variety of applications, CUTLASS provides extensive support for
+mixed-precision computations, providing specialized data-movement and
+multiply-accumulate abstractions for half-precision floating
+point (FP16), BFloat16 (BF16), Tensor Float 32 (TF32),
+single-precision floating point (FP32),
+[FP32 emulation via tensor core instruction](./examples/27_ampere_3xtf32_fast_accurate_tensorop_gemm),
+double-precision floating
+point (FP64) types, integer data types (4b and 8b), and binary data types (1b).
+CUTLASS demonstrates warp-synchronous matrix multiply operations
+targeting the programmable, high-throughput _Tensor Cores_ implemented by
+NVIDIA's Volta, Turing, Ampere, and Hopper architectures.
+
+See the [Quick Start Guide](./media/docs/quickstart.md) to get started quickly.
+
+See the [functionality listing](./media/docs/functionality.md) for the list of operations
+supported at each level of the execution model hierarchy.
+
+CUTLASS 3.0 introduced a new core library, CuTe, to describe and manipulate tensors of threads and data.
+CuTe is a collection of C++ CUDA template abstractions for defining and operating on hierarchically multidimensional layouts of threads and data. CuTe provides `Layout` and `Tensor` objects that compactly package the type, shape, memory space, and layout of data, while performing the complicated indexing for the user. This lets programmers focus on the logical descriptions of their algorithms while CuTe does the mechanical bookkeeping for them. With these tools, we can quickly design, implement, and modify all dense linear algebra operations.
+
+The core abstractions of CuTe are hierarchically multidimensional layouts which can be composed with data arrays to represent tensors. The representation of layouts is powerful enough to represent nearly everything we need to implement efficient dense linear algebra. Layouts can also be combined and manipulated via functional composition, on which we build a large set of common operations such as tiling and partitioning.
+
+CUTLASS 3.0 and beyond adopts CuTe throughout the GEMM hierarchy in its templates. This greatly simplifies the design
+and improves code composability and readability. More documentation specific to CuTe can be found in its [dedicated documentation directory](./media/docs/cute/00_quickstart.md).
+
+In addition to GEMMs, CUTLASS implements high-performance convolution via the implicit GEMM algorithm. Implicit GEMM is the formulation of a convolution operation as a GEMM thereby taking advantage of CUTLASS's modular GEMM pipeline. This allows CUTLASS to build convolutions by reusing highly-optimized GEMM components.
+
+# What's New in CUTLASS 3.6
+
+CUTLASS 3.6.0 is an update to CUTLASS adding:
+
+- [Hopper structured sparse GEMM](./examples/62_hopper_sparse_gemm/62_hopper_sparse_gemm.cu).
+  + [FP16](./test/unit/gemm/device/sm90_sparse_gemm_f16_f16_f32_tensor_op_f32.cu)
+  + [FP8](./test/unit/gemm/device/sm90_sparse_gemm_f8_f8_f32_tensor_op_f32.cu)
+  + [INT8](./test/unit/gemm/device/sm90_sparse_gemm_s8_s8_s32_tensor_op_s32.cu)
+  + [TF32](./test/unit/gemm/device/sm90_sparse_gemm_tf32_tf32_f32_tensor_op_f32.cu)
+- A refactor to the CUTLASS 3.x convolution `kernel::ConvUniversal` [API](./include/cutlass/conv/kernel/sm90_implicit_gemm_tma_warpspecialized.hpp) to bring it in line with `gemm::GemmUniversal`. Now the 3.x convolution API is no longer considered as a beta API.
+- [An improved mixed input GEMM](./examples/55_hopper_mixed_dtype_gemm/README.md) and a [lookup table implementation](./examples/55_hopper_mixed_dtype_gemm/55_hopper_int4_fp8_gemm.cu) for `INT4`x`FP8` scale-only mode.
+- [EVT nodes for Top-K selection and softmax](./include/cutlass/epilogue/fusion/sm90_visitor_topk_softmax.hpp) and [GEMM example using those](./examples/61_hopper_gemm_with_topk_and_softmax/61_hopper_gemm_with_topk_and_softmax.cu).
+- [Programmatic Dependent Launch](./include/cutlass/arch/grid_dependency_control.h) (PDL) that leverages a new Hopper feature to speedup two back-to-back kernels, and its corresponding [documentations](./media/docs/dependent_kernel_launch.md).
+- [A new debugging tool, synclog](./include/cutlass/arch/synclog.hpp), for dumping out all synchronization events from within a kernel to a file. Please see [synclog documentation](./media/docs/utilities.md#debugging-asynchronous-kernels-with-cutlasss-built-in-synclog-tool) for details.
+- A new TMA-enabled [epilogue](./include/cutlass/epilogue/collective/sm90_epilogue_array_tma_warpspecialized.hpp) for grouped GEMM that brings significant performance improvement, as well as its EVT support.
+- A SIMT-enabled pointer-array [epilogue](./include/cutlass/epilogue/collective/sm70_epilogue_vectorized_array.hpp).
+- A new [Ping-Pong kernel schedule for Grouped GEMM](./include/cutlass/gemm/kernel/sm90_gemm_array_tma_warpspecialized_pingpong.hpp) and some other optimizations.
+- [A new instantiation strategy for CUTLASS profiler kernels](./python/cutlass_library/sm90_shapes.py) along with [improved documentation for instantiation level in CUTLASS profiler](./media/docs/profiler.md#instantiating-more-kernels-with-hopper).
+- A new hardware support for comparisons and computations of [`cutlass::bfloat16_t`](./include/cutlass/bfloat16.h)
+- Fixed use of isnan on Windows for [`half_t`](./test/unit/core/functional.cu).
+
+Minimum requirements:
+
+- Architecture: Volta
+- Compiler: Must support at least C++17
+- CUDA Toolkit version: 11.4
+
+Starting from CUTLASS 3.0, CUTLASS removed support for the following:
+
+- Maxwell and Pascal GPU architectures
+- Ubuntu 16.04
+- CUDA 10.2
+- C++ language versions less than 17.
+
+**See the [CHANGELOG](CHANGELOG.md) for a detailed listing of releases and updates.**
+
+# Performance
+
+<p align="center"><img src=media/images/cutlass-3.5.1-gemm-peak-performance.png></p>
+<p align="center"><img src=media/images/cutlass-3.5.1-gemm-peak-performance-fp8.png></p>
+
+CUTLASS primitives are very efficient.  When used to construct device-wide GEMM kernels,
+they exhibit peak performance comparable to cuBLAS for scalar GEMM
+computations. The above figure shows the continual CUTLASS performance improvements 
+on an [NVIDIA H100](https://www.nvidia.com/en-us/data-center/h100/) (NVIDIA Hopper architecture) since
+CUTLASS 3.1.
+CUTLASS 3.5.1 was compiled with the [CUDA 12.5u1 Toolkit](https://developer.nvidia.com/cuda-downloads). 
+Tensor Core operations are implemented using CUDA's 
+[mma](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#warp-level-matrix-instructions-mma) and
+[wgmma](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#asynchronous-warpgroup-level-matrix-instructions) instructions.
+
+<p align="center"><img src=media/images/cutlass-2.9-implicit-gemm-performance.png></p>
+
+When using CUTLASS building blocks to construct device-wide implicit gemm (Fprop, Dgrad, and Wgrad)
+kernels, CUTLASS performance is also comparable to cuDNN when running Resnet-50 layers on an [NVIDIA A100](https://www.nvidia.com/en-us/data-center/a100/)
+as shown in the above figure.  Tensor Core operations are implemented using CUDA's
+[mma instruction](https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#warp-level-matrix-instructions-mma).
+
+# Compatibility
+
+CUTLASS requires a C++17 host compiler and 
+performs best when built with the [**CUDA 12.4 Toolkit**](https://developer.nvidia.com/cuda-downloads).
+It is also compatible with CUDA 11.4, CUDA 11.5, CUDA 11.6, CUDA 11.7, CUDA 11.8, CUDA 12.0, CUDA 12.1, CUDA 12.2.2, CUDA 12.3.1 and CUDA 12.3.2.
+
+## Operating Systems
+We have tested the following environments.
+
+|**Operating System** | **Compiler** |
+|-----------------|----------|
+| Ubuntu 18.04 | GCC 7.5.0  |
+| Ubuntu 20.04 | GCC 10.3.0 |
+| Ubuntu 22.04 | GCC 11.2.0 |
+| Ubuntu 22.04 | Clang 10.0.0 |
+| Ubuntu 22.04 | Clang 14.0.6 |
+| Ubuntu 22.04 | Clang 17.0.6 |
+| Windows 10.0 | Visual Studio 2019 v16.11.27 |
+
+Note: GCC 8.5.0 has known regressions regarding fold expressions and overloaded operators. Using GCC 7.5.0 or (preferred) GCC >= 9 is recommended.
+
+## Hardware
+CUTLASS runs successfully on the following NVIDIA GPUs, and it is expected to be efficient on Volta, Turing, Ampere, Ada, and Hopper architecture based NVIDIA GPUs.
+
+|**GPU**|**CUDA Compute Capability**|**Minimum CUDA Toolkit Required by CUTLASS-3**|
+|---|---|---|
+|NVIDIA V100 Tensor Core GPU            |7.0|11.4|
+|NVIDIA TitanV                          |7.0|11.4|
+|NVIDIA GeForce RTX 2080 TI, 2080, 2070 |7.5|11.4|
+|NVIDIA T4                              |7.5|11.4|
+|NVIDIA A100 Tensor Core GPU            |8.0|11.4|
+|NVIDIA A10                             |8.6|11.4|
+|NVIDIA GeForce RTX 3090                |8.6|11.4|
+|NVIDIA GeForce RTX 4090                |8.9|11.8|
+|NVIDIA L40                             |8.9|11.8|
+|NVIDIA H100 Tensor Core GPU            |9.0|11.8|
+
+## Target Architecture
+
+In general, PTX code generated for one target architecture can be run on future architectures (i.e., it is forward compatible).  However, CUDA 12.0 introduced the concept of "architecture-accelerated features" whose PTX does not have forward compatibility guarantees. Several Hopper PTX instructions fall under this category of architecture-accelerated features, and thus require a `sm_90a` target architecture (note the "a" appended). For more details on this and other architecture-accelerated instructions, please refer to the [CUDA Documentation](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#feature-availability).
+
+The target architecture information is passed on to CUTLASS via the cmake flag `CUTLASS_NVCC_ARCHS`. In order to maximize performance on Hopper GH100, users are required to build CUTLASS with `90a` as the target architecture. If a user accidentally builds a kernel which uses SM90a features (e.g. Hopper Tensor Core Instructions), using the SM90 target (note the lack of "a"), with either CUDA Toolkit 12 or 11.8, the kernel is expected to fail with a runtime error.
+
+```
+cmake .. -DCUTLASS_NVCC_ARCHS="90a" 
+```
+
+Please refer to the [functionality documentation](./media/docs/functionality.md) for details on which kernels require which target architectures.
+
+# Documentation
+
+CUTLASS is described in the following documents and the accompanying
+[Doxygen documentation](https://nvidia.github.io/cutlass).
+
+- [Quick Start Guide](./media/docs/quickstart.md) - build and run CUTLASS
+- [Functionality](./media/docs/functionality.md) - summarizes functionality available in CUTLASS
+- [Efficient GEMM in CUDA](./media/docs/efficient_gemm.md) - describes how GEMM kernels may be implemented efficiently in CUDA
+- [CUTLASS 3.x Design](./media/docs/cutlass_3x_design.md) - describes the CUTLASS 3.x design, its benefits, and how CuTe enables us to write much more composable components
+- [GEMM API 3.x](./media/docs/gemm_api_3x.md) - describes the CUTLASS 3.x GEMM model and C++ template concepts
+- [GEMM API 2.x](./media/docs/gemm_api.md) - describes the CUTLASS 2.x GEMM model and C++ template concepts
+- [Implicit GEMM Convolution](./media/docs/implicit_gemm_convolution.md) - describes 2-D and 3-D convolution in CUTLASS
+- [Code Organization](./media/docs/code_organization.md) - describes the organization and contents of the CUTLASS project
+- [Terminology](./media/docs/terminology.md) - describes terms used in the code
+- [Programming Guidelines](./media/docs/programming_guidelines.md) - guidelines for writing efficient modern CUDA C++
+- [Fundamental types](./media/docs/fundamental_types.md) - describes basic C++ classes used in CUTLASS to represent numeric quantities and arrays
+- [Layouts](./media/docs/layout.md) - describes layouts of matrices and tensors in memory
+- [Tile Iterators](./media/docs/tile_iterator_concept.md) - describes C++ concepts for iterating over tiles of matrices in memory
+- [CUTLASS Profiler](./media/docs/profiler.md) - command-line driven profiling application
+- [CUTLASS Utilities](./media/docs/utilities.md) - additional templates used to facilate rapid development
+- [Dependent kernel launch](./media/docs/dependent_kernel_launch.md) - describes a new feature in Hopper which allows overlapping dependent 
+kernels in the same stream, and how it is used in CUTLASS.
+
+# Resources
+We have also described the structure of an efficient GEMM in our talk at the
+[GPU Technology Conference 2018](http://on-demand.gputechconf.com/gtc/2018/presentation/s8854-cutlass-software-primitives-for-dense-linear-algebra-at-all-levels-and-scales-within-cuda.pdf).
+
+ - [CUTLASS: Software Primitives for Dense Linear Algebra at All Levels and Scales within CUDA](https://www.nvidia.com/en-us/on-demand/session/gtcsiliconvalley2018-s8854/)
+ - [Developing CUDA Kernels to Push Tensor Cores to the Absolute Limit on NVIDIA A100](https://www.nvidia.com/en-us/on-demand/session/gtcsj20-s21745/)
+ - [Accelerating Convolution with Tensor Cores in CUTLASS](https://www.nvidia.com/en-us/on-demand/session/gtcspring21-s31883/)
+ - [Accelerating Backward Data Gradient by Increasing Tensor Core Utilization in CUTLASS](https://www.nvidia.com/en-us/on-demand/session/gtcspring22-s41996/)
+ - [CUTLASS: Python API, Enhancements, and NVIDIA Hopper](https://www.nvidia.com/en-us/on-demand/session/gtcfall22-a41131/)
+
+# Building CUTLASS
+
+CUTLASS is a header-only template library and does not need to be built to be used by other
+projects. Client applications should target CUTLASS's `include/` directory in their include
+paths.
+
+CUTLASS unit tests, examples, and utilities can be build with CMake.
+The minimum version of CMake is given in the [Quickstart guide](./media/docs/quickstart.md).
+Make sure the `CUDACXX` environment  variable points to NVCC in the CUDA Toolkit installed
+on your system.
+
+```bash
+$ export CUDACXX=${CUDA_INSTALL_PATH}/bin/nvcc
+```
+
+Create a build directory within the CUTLASS project, then run CMake. By default CUTLASS will build kernels
+for CUDA architecture versions 5.0, 6.0, 6.1, 7.0, 7.5, 8.0, 8.6, 8.9, and 9.0.
+To reduce compile time you can specify
+the architectures to build CUTLASS for by changing the CMake configuration setting
+`CUTLASS_NVCC_ARCHS`.
+
+```bash
+$ mkdir build && cd build
+
+$ cmake .. -DCUTLASS_NVCC_ARCHS=80               # compiles for NVIDIA's Ampere Architecture
+```
+
+From the `build/` directory, compile and run the CUTLASS unit tests by building the target `test_unit` with make.
+
+The unit tests are organized as several binaries mirroring the top-level namespaces of CUTLASS,
+and they may be executed in parallel via make's `-j` command line argument.
+
+```bash
+$ make test_unit -j
+...
+...
+...
+[----------] Global test environment tear-down
+[==========] 946 tests from 57 test cases ran. (10812 ms total)
+[  PASSED  ] 946 tests.
+```
+
+All tests should pass on supported platforms, though the exact number of tests may vary over time.
+
+
+# Project Structure
+
+CUTLASS is arranged as a header-only library along with Utilities, Tools, Examples, and unit tests. 
+[Doxygen documentation](https://nvidia.github.io/cutlass) provides a complete list of files, classes, 
+and template concepts defined in the CUTLASS project.
+
+A detailed explanation of the source code organization may be found in the 
+[CUTLASS documentation](./media/docs/code_organization.md), but several main components are summarized below.
+
+## CUTLASS Template Library
+
+```
+include/                     # client applications should target this directory in their build's include paths
+
+  cutlass/                   # CUDA Templates for Linear Algebra Subroutines and Solvers - headers only
+
+    arch/                    # direct exposure of architecture features (including instruction-level GEMMs)
+
+    conv/                    # code specialized for convolution
+
+    epilogue/                # code specialized for the epilogue of gemm/convolution
+
+    gemm/                    # code specialized for general matrix product computations
+
+    layout/                  # layout definitions for matrices, tensors, and other mathematical objects in memory
+
+    platform/                # CUDA-capable Standard Library components
+
+    reduction/               # bandwidth-limited reduction kernels that do not fit the "gemm" model
+
+    thread/                  # simt code that can be performed within a CUDA thread
+    
+    transform/               # code specialized for layout, type, and domain transformations
+
+    *                        # core vocabulary types, containers, and basic numeric operations
+
+  cute/                      # CuTe Layout, layout algebra, MMA/Copy atoms, tiled MMA/Copy
+
+    algorithm/               # Definitions of core operations such as copy, gemm, and operations on cute::tuples
+
+    arch/                    # Bare bones PTX wrapper structs for copy and math instructions
+
+    atom/                    # Meta-information either link to or built from arch/ operators
+
+      mma_atom.hpp           # cute::Mma_Atom and cute::TiledMma
+
+      copy_atom.hpp          # cute::Copy_Atom and cute::TiledCopy
+
+      *sm*.hpp               # Arch specific meta-information for copy and math operations
+
+    *                        # Core library types such as Shape, Stride, Layout, Tensor, and associated operations
+
+```
+
+### CUTLASS SDK Examples
+
+[CUTLASS SDK examples](./examples) apply CUTLASS templates to implement basic computations.
+
+### Tools
+
+```
+tools/
+  library/                   # CUTLASS Instance Library - contains instantiations of all supported CUTLASS templates
+    include/
+      cutlass/
+        library/
+
+  profiler/                  # CUTLASS Profiler         - command-line utility for executing operations in the
+                             #                            CUTLASS Library
+  
+  util/                      # CUTLASS Utilities        - contains numerous helper classes for
+    include/                 #                            manging tensors in device memory, reference
+      cutlass/               #                            implementations for GEMM, random initialization
+        util/                #                            of tensors, and I/O.
+```
+
+### Test
+
+The `test/unit/` directory consist of unit tests implemented with Google Test that demonstrate
+basic usage of Core API components and complete tests of the CUTLASS GEMM computations.
+
+Instructions for building and running the Unit tests are described in the [Quickstart guide](./media/docs/quickstart.md).
+
+# Performance Profiling
+
+The `tools/profiler/` directory contains a command-line utility for launching each of the GEMM kernels.
+It can be built as follows:
+
+```bash
+$ make cutlass_profiler -j16
+```
+## Building all GEMM and Convolution kernels (_long_ build times)
+
+By default, only one tile size is instantiated for each data type, math instruction, and layout.
+To instantiate all, set the following environment variable when running CMake from an empty `build/` directory.
+Beware, this results in *tens of thousands* of kernels and long build times. 
+This would also result in a large binary size and on some platforms linker to fail on building the library.
+Therefore, it's highly recommended to generate only a subset of kernels as demonstrated in the sub-section below.
+```bash
+$ cmake .. -DCUTLASS_NVCC_ARCHS=90a -DCUTLASS_LIBRARY_KERNELS=all
+...
+$ make cutlass_profiler -j16
+```
+
+## Building a subset of GEMM and Convolution kernels (_reduced_ build times)
+
+To compile strictly one kernel or a small set of kernels, a comma-delimited list of kernel names with 
+wildcard characters may be used to reduce the set of kernels. The following examples show building exactly one
+or a subset of kernels for NVIDIA Ampere and Turing architecture:
+
+### Building a subset Tensor Core GEMM kernels
+
+To compile a subset of Tensor Core GEMM kernels with FP32 accumulation and FP16 input targeting NVIDIA Ampere and Turing architecture, 
+use the below cmake command line:
+```bash
+$ cmake .. -DCUTLASS_NVCC_ARCHS='75;80' -DCUTLASS_LIBRARY_KERNELS=cutlass_tensorop_s*gemm_f16_*_nt_align8
+...
+$ make cutlass_profiler -j16
+```
+
+Example command line for profiling a subset of Tensor Core GEMM kernels is as follows:
+```bash
+./tools/profiler/cutlass_profiler --kernels=cutlass_tensorop_s*gemm_f16_*_nt_align8 --m=3456 --n=4096 --k=4096
+
+...
+=============================
+  Problem ID: 1
+
+        Provider: CUTLASS
+   OperationKind: gemm
+       Operation: cutlass_tensorop_s1688gemm_f16_256x128_32x2_nt_align8
+
+          Status: Success
+    Verification: ON
+     Disposition: Passed
+
+reference_device: Passed
+          cuBLAS: Passed
+
+       Arguments: --gemm_kind=universal --m=3456 --n=4096 --k=4096 --A=f16:column --B=f16:row --C=f32:column --alpha=1  \
+                  --beta=0 --split_k_slices=1 --batch_count=1 --op_class=tensorop --accum=f32 --cta_m=256 --cta_n=128  \
+                  --cta_k=32 --stages=2 --warps_m=4 --warps_n=2 --warps_k=1 --inst_m=16 --inst_n=8 --inst_k=8 --min_cc=75  \
+                  --max_cc=1024
+
+           Bytes: 118489088  bytes
+           FLOPs: 115992428544  flops
+
+         Runtime: 1.55948  ms
+          Memory: 70.7616 GiB/s
+
+            Math: 74378.8 GFLOP/s
+
+
+
+=============================
+...
+```
+
+### Building one CUDA Core GEMM kernel
+
+To compile one SGEMM kernel targeting NVIDIA Ampere and Turing architecture, use the below cmake command line:
+```bash
+$ cmake .. -DCUTLASS_NVCC_ARCHS='75;80' -DCUTLASS_LIBRARY_KERNELS=cutlass_simt_sgemm_128x128_8x2_nn_align1
+...
+$ make cutlass_profiler -j16
+```
+
+Example command line for profiling single SGEMM CUDA kernel is as follows:
+```bash
+$ ./tools/profiler/cutlass_profiler --kernels=sgemm --m=3456 --n=4096 --k=4096
+
+=============================
+  Problem ID: 1
+
+        Provider: CUTLASS
+   OperationKind: gemm
+       Operation: cutlass_simt_sgemm_128x128_8x2_nn_align1
+
+          Status: Success
+    Verification: ON
+     Disposition: Passed
+
+          cuBLAS: Passed
+
+       Arguments: --m=3456 --n=4096 --k=4096 --A=f32:column --B=f32:column --C=f32:column --alpha=1 --beta=0 --split_k_slices=1  \
+                  --batch_count=1 --op_class=simt --accum=f32 --cta_m=128 --cta_n=128 --cta_k=8 --stages=2 --warps_m=4  \
+                  --warps_n=2 --warps_k=1 --inst_m=1 --inst_n=1 --inst_k=1 --min_cc=50 --max_cc=1024
+
+           Bytes: 180355072  bytes
+           FLOPs: 115992428544  flops
+
+         Runtime: 6.73655  ms
+          Memory: 24.934 GiB/s
+
+            Math: 17218.4 GFLOP/s
+
+=============================
+```
+
+### Building a subset of Tensor Core Convolution kernels
+
+To compile a subset of Tensor core convolution kernels implementing forward propagation (fprop) with FP32 accumulation 
+and FP16 input targeting NVIDIA Ampere and Turing architecture, use the below cmake command line:
+```bash
+$ cmake .. -DCUTLASS_NVCC_ARCHS='75;80' -DCUTLASS_LIBRARY_KERNELS=cutlass_tensorop_s*fprop_optimized_f16
+...
+$ make cutlass_profiler -j16
+```
+
+Example command line for profiling a subset of Tensor Core convolution kernels is as follows:
+
+```bash
+$ ./tools/profiler/cutlass_profiler --kernels=cutlass_tensorop_s*fprop_optimized_f16 --n=8 --h=224 --w=224 --c=128 --k=128 --r=3 --s=3
+
+...
+=============================
+  Problem ID: 1
+
+        Provider: CUTLASS
+   OperationKind: conv2d
+       Operation: cutlass_tensorop_s16816fprop_optimized_f16_128x128_32x5_nhwc
+
+          Status: Success
+    Verification: ON
+     Disposition: Passed
+
+reference_device: Passed
+
+       Arguments: --conv_kind=fprop --n=8 --h=224 --w=224 --c=128 --k=128 --r=3 --s=3 --p=224 --q=224 --pad_h=1 --pad_w=1  \
+                  --stride_h=1 --stride_w=1 --dilation_h=1 --dilation_w=1 --Activation=f16:nhwc --Filter=f16:nhwc --Output=f32:nhwc  \
+                  --conv_mode=cross --iterator_algorithm=optimized --alpha=1 --beta=0 --split_k_mode=serial --split_k_slices=1  \
+                  --eq_gemm_provider=none --op_class=tensorop --accum=f32 --cta_m=128 --cta_n=128 --cta_k=32 --stages=5  \
+                  --warps_m=2 --warps_n=2 --warps_k=1 --inst_m=16 --inst_n=8 --inst_k=16 --min_cc=80 --max_cc=1024
+
+           Bytes: 1130659840  bytes
+           FLOPs: 118482796544  flops
+
+         Runtime: 0.711496  ms
+          Memory: 1479.99 GiB/s
+
+            Math: 166526 GFLOP/s
+
+=============================
+...
+```
+
+
+### Building one Convolution CUDA kernel
+
+To compile and run one CUDA Core convolution kernel implementing forward propagation (fprop) with F32 accumulation 
+and FP32 input targeting NVIDIA Ampere and Turing architecture, use the below cmake command line:
+```bash
+$ cmake .. -DCUTLASS_NVCC_ARCHS='75;80' -DCUTLASS_LIBRARY_KERNELS=cutlass_simt_sfprop_optimized_128x128_8x2_nhwc
+...
+$ make cutlass_profiler -j16
+```
+
+Example command line for profiling one CUDA Core convolution kernel:
+
+```bash
+$ ./tools/profiler/cutlass_profiler --kernels=cutlass_simt_sfprop_optimized_128x128_8x2_nhwc --n=8 --h=224 --w=224 --c=128 --k=128 --r=3 --s=3
+
+
+=============================
+  Problem ID: 1
+
+        Provider: CUTLASS
+   OperationKind: conv2d
+       Operation: cutlass_simt_sfprop_optimized_128x128_8x2_nhwc
+
+          Status: Success
+    Verification: ON
+     Disposition: Passed
+
+reference_device: Passed
+
+       Arguments: --conv_kind=fprop --n=8 --h=224 --w=224 --c=128 --k=128 --r=3 --s=3 --p=224 --q=224 --pad_h=1 --pad_w=1  \
+                  --stride_h=1 --stride_w=1 --dilation_h=1 --dilation_w=1 --Activation=f32:nhwc --Filter=f32:nhwc --Output=f32:nhwc  \
+                  --conv_mode=cross --iterator_algorithm=optimized --alpha=1 --beta=0 --split_k_mode=serial --split_k_slices=1  \
+                  --eq_gemm_provider=none --op_class=simt --accum=f32 --cta_m=128 --cta_n=128 --cta_k=8 --stages=2 --warps_m=4  \
+                  --warps_n=2 --warps_k=1 --inst_m=1 --inst_n=1 --inst_k=1 --min_cc=50 --max_cc=1024
+
+           Bytes: 2055798784  bytes
+           FLOPs: 118482796544  flops
+
+         Runtime: 7.34266  ms
+          Memory: 260.752 GiB/s
+
+            Math: 16136.2 GFLOP/s
+
+
+=============================
+
+```
+
+## More Details on Compiling CUTLASS Kernels and CUTLASS Profiler
+- Please follow the links for more CMake examples on selectively compiling CUTLASS kernels:
+  - [GEMM CMake Examples](./media/docs/quickstart.md#gemm-cmake-examples) 
+  - [Implicit GEMM convolution CMake Examples](./media/docs/quickstart.md#convolution-cmake-examples)
+- [Further details about the CUTLASS Profiler are described here.](./media/docs/profiler.md)
+
+
+# About
+
+CUTLASS is released by NVIDIA Corporation as Open Source software under the 
+[3-clause "New" BSD license](LICENSE.txt).
+
+# Contributors
+
+The official list of CUTLASS developers and contributors is available here: [CONTRIBUTORS](CONTRIBUTORS.md).
+
+# Copyright
+
+Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+SPDX-License-Identifier: BSD-3-Clause
+
+```
+  Redistribution and use in source and binary forms, with or without
+  modification, are permitted provided that the following conditions are met:
+
+  1. Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+  2. Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+  3. Neither the name of the copyright holder nor the names of its
+  contributors may be used to endorse or promote products derived from
+  this software without specific prior written permission.
+
+  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+  AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+  IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+  DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+  DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+  SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+  CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+  OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+  OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+```

crates/vendor/candle/candle-flash-attn/cutlass/bin2hex.cmake

@@ -0,0 +1,54 @@
+# Copyright (c) 2019 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# A small utility function which generates a C-header from an input file
+function(FILE_TO_C_STRING FILENAME VARIABLE_NAME OUTPUT_STRING ZERO_TERMINATED)
+  FILE(READ "${FILENAME}" HEX_INPUT HEX)
+  if (${ZERO_TERMINATED})
+    string(APPEND HEX_INPUT "00")
+  endif()
+
+  string(REGEX REPLACE "(....)" "\\1\n" HEX_OUTPUT ${HEX_INPUT})
+  string(REGEX REPLACE "([0-9a-f][0-9a-f])" "char(0x\\1)," HEX_OUTPUT ${HEX_OUTPUT})
+
+  set(HEX_OUTPUT "static char const ${VARIABLE_NAME}[] = {\n  ${HEX_OUTPUT}\n};\n")
+
+  set(${OUTPUT_STRING} "${HEX_OUTPUT}" PARENT_SCOPE)
+endfunction()
+
+# message("Create header file for ${FILE_IN}")
+# message("Create header file for ${FILE_OUT}")
+file_to_c_string(${FILE_IN} ${VARIABLE_NAME} OUTPUT_STRING ZERO_TERMINATED)
+
+set(RESULT "#pragma once\n")
+string(APPEND RESULT "namespace cutlass {\n")
+string(APPEND RESULT "namespace nvrtc {\n")
+string(APPEND RESULT "${OUTPUT_STRING}")
+string(APPEND RESULT "} // namespace nvrtc\n")
+string(APPEND RESULT "} // namespace cutlass\n")
+file(WRITE "${FILE_OUT}" "${RESULT}")

crates/vendor/candle/candle-flash-attn/cutlass/cmake/CTestTestfile.configure.cmake

@@ -0,0 +1,52 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Generated file
+
+set(TEST_SETS_SUPPORTED @TEST_SETS_SUPPORTED@)
+
+if (NOT DEFINED ENV{CUTLASS_TEST_SETS})
+  set(ENV{CUTLASS_TEST_SETS} @CUTLASS_DEFAULT_ACTIVE_TEST_SETS@)
+endif()
+
+foreach(TEST_SET_REQUESTED IN ITEMS $ENV{CUTLASS_TEST_SETS})
+  if (NOT TEST_SET_REQUESTED IN_LIST TEST_SETS_SUPPORTED) 
+    message(STATUS "Skipping tests for @TEST_EXE_PATH@ as ${TEST_SET_REQUESTED} is not in the set of [${TEST_SETS_SUPPORTED}].")
+    return()
+  endif()
+endforeach()
+
+set(TEST_EXE_PATH @TEST_EXE_PATH@)
+set(TEST_EXE_WORKING_DIRECTORY @TEST_EXE_WORKING_DIRECTORY@)
+set(CUTLASS_USE_EXTENDED_ADD_TEST_FORMAT @TEST_USE_EXTENDED_FORMAT@)
+
+if (DEFINED ENV{CUTLASS_TEST_EXECUTION_ENVIRONMENT})
+  set(_CUTLASS_TEST_EXECUTION_ENVIRONMENT $ENV{CUTLASS_TEST_EXECUTION_ENVIRONMENT})
+else()
+  set(_CUTLASS_TEST_EXECUTION_ENVIRONMENT @CUTLASS_TEST_EXECUTION_ENVIRONMENT@)
+endif()

crates/vendor/candle/candle-flash-attn/cutlass/cmake/CTestTestfile.test.configure.cmake

@@ -0,0 +1,43 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+if (CUTLASS_USE_EXTENDED_ADD_TEST_FORMAT)
+  # The longform/extended format allows generator expressions to be
+  # expanded property and is useful in contexts where the files need
+  # to be immediately included into being-processed cmake code.
+  add_test(NAME @TESTCASE_NAME@ COMMAND ${_CUTLASS_TEST_EXECUTION_ENVIRONMENT} "${TEST_EXE_PATH}" @TEST_COMMAND_OPTIONS@)
+else()
+  add_test(@TESTCASE_NAME@ ${_CUTLASS_TEST_EXECUTION_ENVIRONMENT} "${TEST_EXE_PATH}" @TEST_COMMAND_OPTIONS@)
+endif()
+
+if (TEST_EXE_WORKING_DIRECTORY)
+  set_tests_properties(@TESTCASE_NAME@ PROPERTIES WORKING_DIRECTORY "${TEST_EXE_WORKING_DIRECTORY}")
+endif()
+
+set_tests_properties(@TESTCASE_NAME@ PROPERTIES DISABLED @__DISABLE_TESTS@)
+

crates/vendor/candle/candle-flash-attn/cutlass/cmake/NvidiaCutlassConfig.cmake.in

@@ -0,0 +1,9 @@
+get_filename_component(NvidiaCutlass_CMAKE_DIR "${CMAKE_CURRENT_LIST_FILE}" PATH)
+
+include(CMakeFindDependencyMacro)
+
+if(TARGET nvidia::cutlass::CUTLASS)
+  return()
+endif()
+
+include("${NvidiaCutlass_CMAKE_DIR}/NvidiaCutlassTargets.cmake")

crates/vendor/candle/candle-flash-attn/cutlass/cmake/NvidiaCutlassPackageConfig.cmake

@@ -0,0 +1,42 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set(CPACK_PACKAGE_NAME NvidiaCutlass)
+set(CPACK_PACKAGE_VENDOR NVIDIA)
+set(CPACK_PACKAGE_CONTACT info@nvidia.com)
+set(CPACK_PACKAGE_DESCRIPTION_SUMMARY "CUTLASS CUDA C++ Template Linear Algebra Library")
+set(CPACK_PACKAGE_INSTALL_DIRECTORY ${CPACK_PACKAGE_NAME})
+set(CPACK_PACKAGE_VERSION_MAJOR ${PROJECT_VERSION_MAJOR})
+set(CPACK_PACKAGE_VERSION_MINOR ${PROJECT_VERSION_MINOR})
+set(CPACK_PACKAGE_VERSION_PATCH ${PROJECT_VERSION_PATCH})
+set(CPACK_VERBATIM_VARIABLES YES)
+# set(CPACK_PACKAGE_DESCRIPTION_FILE ${CMAKE_CURRENT_LIST_DIR}/Description.txt)
+# set(CPACK_RESOURCE_FILE_WELCOME ${CMAKE_CURRENT_LIST_DIR}/Welcome.txt)
+# set(CPACK_RESOURCE_FILE_LICENSE ${CMAKE_CURRENT_LIST_DIR}/License.txt)
+# set(CPACK_RESOURCE_FILE_README ${CMAKE_CURRENT_LIST_DIR}/Readme.txt)
+include(CPack)

crates/vendor/candle/candle-flash-attn/cutlass/cmake/googletest.cmake

@@ -0,0 +1,52 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include(FetchContent)
+
+set(GOOGLETEST_DIR "" CACHE STRING "Location of local GoogleTest repo to build against")
+
+if(GOOGLETEST_DIR)
+  set(FETCHCONTENT_SOURCE_DIR_GOOGLETEST ${GOOGLETEST_DIR} CACHE STRING "GoogleTest source directory override")
+endif()
+
+set(GTEST_REPOSITORY "https://github.com/google/googletest.git" CACHE STRING "GoogleTest repo to fetch")
+FetchContent_Declare(
+  googletest
+  GIT_REPOSITORY ${GTEST_REPOSITORY}
+  GIT_TAG        v1.14.0
+  )
+
+FetchContent_GetProperties(googletest)
+
+if(NOT googletest_POPULATED)
+  FetchContent_Populate(googletest)
+  if (MSVC)
+    set(gtest_force_shared_crt ON CACHE BOOL "" FORCE)
+  endif()
+  add_subdirectory(${googletest_SOURCE_DIR} ${googletest_BINARY_DIR} EXCLUDE_FROM_ALL)
+endif()

crates/vendor/candle/candle-flash-attn/cutlass/cmake/nop.cu

@@ -0,0 +1,49 @@
+/***************************************************************************************************
+ * Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: BSD-3-Clause
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+ * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+ * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ **************************************************************************************************/
+/*! \file
+    \brief Basic CUDA file for testing compiler flags.
+*/
+
+__device__ int inner()
+{
+  return -1;
+}
+
+__global__ void test()
+{
+  inner();
+}
+
+int main()
+{
+  test<<<1,1>>>();
+  return 0;
+}

crates/vendor/candle/candle-flash-attn/cutlass/cmake/version_extended.h.in

@@ -0,0 +1,34 @@
+/***************************************************************************************************
+ * Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: BSD-3-Clause
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials provided with the distribution.
+ *
+ * 3. Neither the name of the copyright holder nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+ * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+ * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ **************************************************************************************************/
+#pragma once
+
+#define CUTLASS_BUILD @CUTLASS_VERSION_BUILD@
+#define CUTLASS_REVISION "@CUTLASS_REVISION@"

crates/vendor/candle/candle-flash-attn/cutlass/cuBLAS.cmake

@@ -0,0 +1,152 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+message(STATUS "Configuring cublas ...")
+
+if((DEFINED CUTLASS_ENABLE_CUBLAS AND NOT CUTLASS_ENABLE_CUBLAS) OR
+   (DEFINED CUBLAS_ENABLED AND NOT CUBLAS_ENABLED))
+  
+  # Don't add cuBLAS if it's defined and false, assume it's not found.
+
+  set(CUBLAS_FOUND OFF)
+  message(STATUS "cuBLAS Disabled.")
+
+elseif(NOT TARGET cublas)
+ 
+  find_path(
+    _CUBLAS_INCLUDE_DIR
+    NAMES cublas_v2.h
+    HINTS
+      ${CUBLAS_INCLUDE_PATH}
+      ENV CUBLAS_INCLUDE_PATH
+      ${CUBLAS_PATH}
+      ENV CUBLAS_PATH
+      ${CUDA_TOOLKIT_ROOT_DIR}
+    PATH_SUFFIXES
+      include
+    )
+
+  find_library(
+    _CUBLAS_LIBRARY
+    NAMES cublas
+    HINTS
+      ${CUBLAS_LIBRARY_PATH}
+      ENV CUBLAS_LIBRARY_PATH
+      ${_CUBLAS_INCLUDE_DIR}/..
+      ${CUBLAS_PATH}
+      ENV CUBLAS_PATH
+      ${CUDA_TOOLKIT_ROOT_DIR}
+    PATH_SUFFIXES
+      lib64
+      lib/x64
+      lib
+    )
+
+  if(_CUBLAS_INCLUDE_DIR AND _CUBLAS_LIBRARY)
+
+    message(STATUS "cuBLAS: ${_CUBLAS_LIBRARY}")
+    message(STATUS "cuBLAS: ${_CUBLAS_INCLUDE_DIR}")
+    
+    set(CUBLAS_FOUND ON CACHE INTERNAL "cublas Library Found")
+    set(CUBLAS_LIBRARY ${_CUBLAS_LIBRARY})
+    set(CUBLAS_INCLUDE_DIR ${_CUBLAS_INCLUDE_DIR})
+
+  else()
+
+    message(STATUS "cublas not found.")
+    set(CUBLAS_FOUND OFF CACHE INTERNAL "cublas Library Found")
+
+  endif()
+
+endif()
+
+set(CUTLASS_ENABLE_CUBLAS ${CUBLAS_FOUND} CACHE BOOL "Enable CUTLASS to build with cuBLAS library.")
+
+if(CUTLASS_ENABLE_CUBLAS AND NOT CUBLAS_FOUND)
+  message(FATAL_ERROR "CUTLASS_ENABLE_CUBLAS enabled but cuBLAS library could not be found.")
+endif()
+
+if(CUTLASS_ENABLE_CUBLAS AND NOT TARGET cublas)
+
+  if(WIN32)
+    add_library(cublas STATIC IMPORTED GLOBAL)
+  else()
+    add_library(cublas SHARED IMPORTED GLOBAL)
+  endif()
+
+  add_library(nvidia::cublas ALIAS cublas)
+
+  set_property(
+    TARGET cublas
+    PROPERTY IMPORTED_LOCATION
+    ${CUBLAS_LIBRARY})
+    
+  target_include_directories(
+    cublas
+    INTERFACE
+    $<INSTALL_INTERFACE:include>
+    $<BUILD_INTERFACE:${CUBLAS_INCLUDE_DIR}>)
+
+  find_library(
+    _CUBLASLT_LIBRARY
+    NAMES cublasLt
+    HINTS
+      ${CUBLAS_LIBRARY_PATH}
+      ENV CUBLAS_LIBRARY_PATH
+      ${_CUBLAS_INCLUDE_DIR}/..
+      ${CUBLAS_PATH}
+      ENV CUBLAS_PATH
+      ${CUDA_TOOLKIT_ROOT_DIR}
+    PATH_SUFFIXES
+      lib64
+      lib/x64
+      lib
+    )
+
+  if(_CUBLASLT_LIBRARY AND NOT TARGET cublasLt)
+
+    if(WIN32)
+      add_library(cublasLt STATIC IMPORTED GLOBAL)
+    else()
+      add_library(cublasLt SHARED IMPORTED GLOBAL)
+    endif()
+    
+    set_property(
+      TARGET cublasLt
+      PROPERTY IMPORTED_LOCATION
+      ${_CUBLASLT_LIBRARY})
+  
+    add_library(nvidia::cublasLt ALIAS cublasLt)
+
+    target_link_libraries(cublas INTERFACE cublasLt)
+
+  endif()
+
+endif()
+
+message(STATUS "Configuring cuBLAS ... done.")

crates/vendor/candle/candle-flash-attn/cutlass/cuDNN.cmake

@@ -0,0 +1,112 @@
+# Copyright (c) 2017 - 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# 1. Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# 3. Neither the name of the copyright holder nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+if(DEFINED CUDNN_ENABLED)
+    set(CUTLASS_ENABLE_CUDNN ${CUDNN_ENABLED} CACHE BOOL "Enable CUTLASS to build with cuDNN library.")
+endif()
+
+if(DEFINED CUTLASS_ENABLE_CUDNN AND NOT CUTLASS_ENABLE_CUDNN)
+  return()
+endif()
+  
+message(STATUS "Configuring cuDNN ...")
+
+find_path(
+    _CUDNN_INCLUDE_DIR cudnn.h
+    PATHS
+    ${CUDA_TOOLKIT_ROOT_DIR}/include
+    $ENV{CUDNN_PATH}/include
+    $ENV{CUDA_PATH}/include
+    ${CUDNN_PATH}/include
+    /usr/include)
+
+find_library(
+    _CUDNN_LIBRARY cudnn
+    HINTS
+    ${CUDA_TOOLKIT_ROOT_DIR}/lib64
+    ${CUDA_TOOLKIT_ROOT_DIR}/lib/x64
+    ${CUDA_TOOLKIT_ROOT_DIR}/lib
+    $ENV{CUDNN_PATH}/lib64
+    $ENV{CUDNN_PATH}/lib/x64
+    $ENV{CUDNN_PATH}/lib
+    $ENV{CUDA_PATH}/lib64
+    $ENV{CUDA_PATH}/lib/x64
+    $ENV{CUDA_PATH}/lib
+    ${CUDNN_PATH}/lib64
+    ${CUDNN_PATH}/lib/x64
+    ${CUDNN_PATH}/lib
+    /usr/lib/x86_64-linux-gnu
+    /usr/lib)
+
+if(_CUDNN_INCLUDE_DIR AND _CUDNN_LIBRARY)
+
+    message(STATUS "cuDNN: ${_CUDNN_LIBRARY}")
+    message(STATUS "cuDNN: ${_CUDNN_INCLUDE_DIR}")
+    
+    set(CUDNN_FOUND ON CACHE INTERNAL "cuDNN Library Found")
+
+else()
+
+    message(STATUS "cuDNN not found.")
+    set(CUDNN_FOUND OFF CACHE INTERNAL "cuDNN Library Found")
+
+endif()
+
+set(CUTLASS_ENABLE_CUDNN ${CUDNN_FOUND} CACHE BOOL "Enable CUTLASS to build with cuDNN library.")
+
+if (CUTLASS_ENABLE_CUDNN AND NOT TARGET cudnn)
+
+  set(CUDNN_INCLUDE_DIR ${_CUDNN_INCLUDE_DIR})
+  set(CUDNN_LIBRARY ${_CUDNN_LIBRARY})
+
+  if(WIN32)
+    add_library(cudnn STATIC IMPORTED GLOBAL)
+  else()
+    add_library(cudnn SHARED IMPORTED GLOBAL)
+  endif()
+
+  add_library(nvidia::cudnn ALIAS cudnn)
+
+  set_property(
+    TARGET cudnn
+    PROPERTY IMPORTED_LOCATION
+    ${CUDNN_LIBRARY})
+    
+  target_include_directories(
+    cudnn
+    INTERFACE
+    $<INSTALL_INTERFACE:include>
+    $<BUILD_INTERFACE:${CUDNN_INCLUDE_DIR}>)
+
+endif()
+
+if(CUTLASS_ENABLE_CUDNN AND NOT CUDNN_FOUND)
+  message(FATAL_ERROR "CUTLASS_ENABLE_CUDNN enabled but cuDNN library could not be found.")
+endif()
+
+message(STATUS "Configuring cuDNN ... done.")

crates/vendor/candle/candle-flash-attn/cutlass/docs/_config.yml

@@ -0,0 +1 @@
+theme: jekyll-theme-minimal

crates/vendor/candle/candle-flash-attn/cutlass/docs/aligned__buffer_8h.html

@@ -0,0 +1,145 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: aligned_buffer.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">aligned_buffer.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>AlignedBuffer is a container for trivially copyable elements suitable for use in unions and shared memory.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &quot;<a class="el" href="cutlass_8h_source.html">cutlass/cutlass.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="array_8h_source.html">cutlass/array.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for aligned_buffer.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="aligned__buffer_8h__incl.png" border="0" usemap="#aligned__buffer_8h" alt=""/></div>
+<map name="aligned__buffer_8h" id="aligned__buffer_8h">
+</map>
+</div>
+</div><div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="aligned__buffer_8h__dep__incl.png" border="0" usemap="#aligned__buffer_8hdep" alt=""/></div>
+<map name="aligned__buffer_8hdep" id="aligned__buffer_8hdep">
+</map>
+</div>
+</div>
+<p><a href="aligned__buffer_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1AlignedBuffer.html">cutlass::AlignedBuffer&lt; T, N, Align &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Modifies semantics of cutlass::Array&lt;&gt; to provide guaranteed alignment.  <a href="structcutlass_1_1AlignedBuffer.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/aligned__buffer_8h__dep__incl.md5

@@ -0,0 +1 @@
+6cbc6b81ede44b5f08afd4f4519d56d1

crates/vendor/candle/candle-flash-attn/cutlass/docs/aligned__buffer_8h__incl.md5

@@ -0,0 +1 @@
+b26c62930ff7668b89f2ee6624e0be3a

crates/vendor/candle/candle-flash-attn/cutlass/docs/annotated.html

@@ -0,0 +1,867 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: Class List</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li class="current"><a href="annotated.html"><span>Classes</span></a></li>
+      <li><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li class="current"><a href="annotated.html"><span>Class&#160;List</span></a></li>
+      <li><a href="classes.html"><span>Class&#160;Index</span></a></li>
+      <li><a href="inherits.html"><span>Class&#160;Hierarchy</span></a></li>
+      <li><a href="functions.html"><span>Class&#160;Members</span></a></li>
+    </ul>
+  </div>
+</div><!-- top -->
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div class="header">
+  <div class="headertitle">
+<div class="title">Class List</div>  </div>
+</div><!--header-->
+<div class="contents">
+<div class="textblock">Here are the classes, structs, unions and interfaces with brief descriptions:</div><div class="directory">
+<div class="levels">[detail level <span onclick="javascript:toggleLevel(1);">1</span><span onclick="javascript:toggleLevel(2);">2</span><span onclick="javascript:toggleLevel(3);">3</span><span onclick="javascript:toggleLevel(4);">4</span><span onclick="javascript:toggleLevel(5);">5</span><span onclick="javascript:toggleLevel(6);">6</span>]</div><table class="directory">
+<tr id="row_0_" class="even"><td class="entry"><span style="width:0px;display:inline-block;">&#160;</span><span id="arr_0_" class="arrow" onclick="toggleFolder('0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass.html" target="_self">cutlass</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_0_" class="arrow" onclick="toggleFolder('0_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1arch.html" target="_self">arch</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma.html" target="_self">Mma</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_30fa42e1ad201df010637cd22fc070a1.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; double &gt;, LayoutA, complex&lt; double &gt;, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_48b3a43bc03fff93a111ac01abe7e40d.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; double &gt;, LayoutA, double, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_76f9d24016e1b4167b16f4d7628c9546.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; float &gt;, LayoutA, complex&lt; float &gt;, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_f1c9d2ee842455cd0c5b71d56108d468.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; float &gt;, LayoutA, float, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_070b94670e040ed5855e5b42d5ca8a443.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, double, LayoutA, complex&lt; double &gt;, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_0aa57e6a2e6b5da37d10688bf99419a23.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, double, LayoutA, double, LayoutB, double, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01ElementAb6e65b2cf5ede7f41cb070a767158dee.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, Operator &gt;</a></td><td class="desc">Matrix multiply-add operation - specialized for 1x1x1x1 matrix multiply operation </td></tr>
+<tr id="row_0_0_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_00e3e12e263df6506b8cf06c3f4d478b8e.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, float, LayoutA, complex&lt; float &gt;, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_004bb3fd76ca2af7b3210676fa9644d95b.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, float, LayoutA, float, LayoutB, float, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01half__t_4f30ee91f7bb3844ff7579c68d078818.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, float, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01int_00_00b2dff9ce8caad9aff5bc6a355539161.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, int, LayoutA, int, LayoutB, int, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_012_01_4_00_011_00_01int16__t8c4bac365710598317a69c489f7239db.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 2 &gt;, 1, int16_t, layout::RowMajor, int16_t, layout::ColumnMajor, int, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_014_01_4_00_011_00_01int8__t_a1ef6624fc8c10126f17f4ee88283d72.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 1, 4 &gt;, 1, int8_t, LayoutA, int8_t, LayoutB, int, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_012_00_011_01_4_00_011_00_01half__t_f3dc2e59f857ada163d1e0781ea8f391.html" target="_self">Mma&lt; gemm::GemmShape&lt; 1, 2, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_0116_00_0116_00_014_01_4_00_0132_00_01half_0bcc4d05f9811035f08cc1b7f0154a4d.html" target="_self">Mma&lt; gemm::GemmShape&lt; 16, 16, 4 &gt;, 32, half_t, LayoutA, half_t, LayoutB, ElementC, LayoutC, Operator &gt;</a></td><td class="desc">Matrix multiply-add operation specialized for the entire warp </td></tr>
+<tr id="row_0_0_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_0116_00_018_00_018_01_4_00_0132_00_01half__02a3f19a78995f97d793a668e0e4d4f0.html" target="_self">Mma&lt; gemm::GemmShape&lt; 16, 8, 8 &gt;, 32, half_t, layout::RowMajor, half_t, layout::ColumnMajor, float, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F32 = F16 * F16 + F32 </td></tr>
+<tr id="row_0_0_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_0116_00_018_00_018_01_4_00_0132_00_01half__96363097c47b056f0ca1911afd7f8b7a.html" target="_self">Mma&lt; gemm::GemmShape&lt; 16, 8, 8 &gt;, 32, half_t, layout::RowMajor, half_t, layout::ColumnMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation - F16 = F16 * F16 + F16 </td></tr>
+<tr id="row_0_0_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_011_00_011_01_4_00_011_00_01half__t_8cf78649807b93684f3d431bfa34ee28.html" target="_self">Mma&lt; gemm::GemmShape&lt; 2, 1, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, half_t, LayoutC, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_ccde11d1bbbdab3702772ce44eb9729a.html" target="_self">Mma&lt; gemm::GemmShape&lt; 2, 2, 1 &gt;, 1, half_t, layout::ColumnMajor, half_t, layout::RowMajor, half_t, layout::ColumnMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_c07cc6439298fa5486a719e577be2538.html" target="_self">Mma&lt; gemm::GemmShape&lt; 2, 2, 1 &gt;, 1, half_t, layout::ColumnMajor, half_t, layout::RowMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_21_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_01128_01_4_00_0132_00_01uint15918972b95027764b3a849b03075ed2b.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 128 &gt;, 32, uint1b_t, layout::RowMajor, uint1b_t, layout::ColumnMajor, int, layout::RowMajor, OpXorPopc &gt;</a></td><td class="desc">Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_22_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01int8__927179f46017ea5f58f859f1196c4829.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, int8_t, layout::RowMajor, int8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * S8 + S32 </td></tr>
+<tr id="row_0_0_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01int8__8ebae0cbdf333fddfe5c24d35ebe8e02.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, int8_t, layout::RowMajor, int8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * S8 + S32 </td></tr>
+<tr id="row_0_0_24_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01int8__5299c9c90c8f2f521be0c8cec1c3eb08.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, int8_t, layout::RowMajor, uint8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * U8 + S32 </td></tr>
+<tr id="row_0_0_25_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01int8__f083347e265b1e9eea5572d86ddb6bf9.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, int8_t, layout::RowMajor, uint8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * U8 + S32 </td></tr>
+<tr id="row_0_0_26_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01uint8_a62aa63a212985df306fb27e8a50aeae.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, uint8_t, layout::RowMajor, int8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U8 * S8 + S32 </td></tr>
+<tr id="row_0_0_27_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01uint8_ab741d81fdc991345cb9e43c29fca573.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, uint8_t, layout::RowMajor, int8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U8 * S8 + S32 </td></tr>
+<tr id="row_0_0_28_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01uint8_5221708cec5828d35db1d1c47cb4964e.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, uint8_t, layout::RowMajor, uint8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * U8 + S32 </td></tr>
+<tr id="row_0_0_29_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0116_01_4_00_0132_00_01uint8_bef0c048bc0f8ba2d875cb7ab26d363b.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 16 &gt;, 32, uint8_t, layout::RowMajor, uint8_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S8 * U8 + S32 </td></tr>
+<tr id="row_0_0_30_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01int4b_6e513ccbc44ae7909a60d93b9b5435b3.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, int4b_t, layout::RowMajor, int4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S4 * S4 + S32 </td></tr>
+<tr id="row_0_0_31_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01int4b_0ee08a4520882d24ba9026879265e892.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, int4b_t, layout::RowMajor, int4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S4 * S4 + S32 </td></tr>
+<tr id="row_0_0_32_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01int4b_4746fc55e614df0016c518d3fda2677e.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S4 * U4 + S32 </td></tr>
+<tr id="row_0_0_33_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01int4b_546e9ec6de6a5970b326da6f6280f1d4.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, int4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = S4 * U4 + S32 </td></tr>
+<tr id="row_0_0_34_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01uint4b03e3b50dbcb30d0d1ac062f3a9d5abef.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, uint4b_t, layout::RowMajor, int4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U4 * S4 + S32 </td></tr>
+<tr id="row_0_0_35_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01uint4b6d968039dde5c9f062ab15f90a8049fe.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, uint4b_t, layout::RowMajor, int4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U4 * S4 + S32 </td></tr>
+<tr id="row_0_0_36_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01uint4bc4b6ba004e25c44bfd9266c61f937dfb.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, uint4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U4 * U4 + S32 </td></tr>
+<tr id="row_0_0_37_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_0132_01_4_00_0132_00_01uint4b451d5cf5d7e8cbbe476afe3dab5c09b2.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 32 &gt;, 32, uint4b_t, layout::RowMajor, uint4b_t, layout::ColumnMajor, int, layout::RowMajor, OpMultiplyAddSaturate &gt;</a></td><td class="desc">Matrix multiply-add operation: S32 = U4 * U4 + S32 </td></tr>
+<tr id="row_0_0_38_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_b0242d7a01097510effbc4718040d3e5.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::ColumnMajor, half_t, layout::ColumnMajor, float, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F32 = F16 * F16 + F32 </td></tr>
+<tr id="row_0_0_39_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_c7f88bfd32a544fba8111d2dcadeab11.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::ColumnMajor, half_t, layout::ColumnMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F16 = F16 * F16 + F16 </td></tr>
+<tr id="row_0_0_40_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_44a3b2a8df88a2b067f1284515cb5371.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::ColumnMajor, half_t, layout::RowMajor, float, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F32 = F16 * F16 + F32 </td></tr>
+<tr id="row_0_0_41_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_4b7308177b308a272c1889fbe9670275.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::ColumnMajor, half_t, layout::RowMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F16 = F16 * F16 + F16 </td></tr>
+<tr id="row_0_0_42_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_5a9888862cebd333ecaf11f7262f77d4.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::RowMajor, half_t, layout::ColumnMajor, float, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F32 = F16 * F16 + F32 </td></tr>
+<tr id="row_0_0_43_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_31defda8ea2b7d855642ffd77da1a411.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::RowMajor, half_t, layout::ColumnMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F16 = F16 * F16 + F16 </td></tr>
+<tr id="row_0_0_44_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_839a7c8bb938d1661f4611e68f85d8cb.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::RowMajor, half_t, layout::RowMajor, float, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F32 = F16 * F16 + F32 </td></tr>
+<tr id="row_0_0_45_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_018_00_018_00_014_01_4_00_018_00_01half__t_73d9802d6b944a5299bc255887db6bbc.html" target="_self">Mma&lt; gemm::GemmShape&lt; 8, 8, 4 &gt;, 8, half_t, layout::RowMajor, half_t, layout::RowMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td><td class="desc">Matrix multiply-add operation: F16 = F16 * F16 + F16 </td></tr>
+<tr id="row_0_0_46_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1PtxWmma.html" target="_self">PtxWmma</a></td><td class="desc">WMMA Matrix multiply-add operation </td></tr>
+<tr id="row_0_0_47_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1PtxWmmaLoadA.html" target="_self">PtxWmmaLoadA</a></td><td class="desc">WMMA PTX string load for A, B, and C matrices </td></tr>
+<tr id="row_0_0_48_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1PtxWmmaLoadB.html" target="_self">PtxWmmaLoadB</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_49_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1PtxWmmaLoadC.html" target="_self">PtxWmmaLoadC</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_50_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1PtxWmmaStoreD.html" target="_self">PtxWmmaStoreD</a></td><td class="desc">WMMA store for matrix D </td></tr>
+<tr id="row_0_0_51_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm50.html" target="_self">Sm50</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_52_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm60.html" target="_self">Sm60</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_53_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm61.html" target="_self">Sm61</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_54_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm70.html" target="_self">Sm70</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_55_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm72.html" target="_self">Sm72</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_56_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Sm75.html" target="_self">Sm75</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_57_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Wmma_3_01Shape___00_01cutlass_1_1half__t_00_01LayoutA___00_01cutlass_1_84e30c8cc93eeb7ca02f651bd16d4c38.html" target="_self">Wmma&lt; Shape_, cutlass::half_t, LayoutA_, cutlass::half_t, LayoutB_, ElementC_, LayoutC_, cutlass::arch::OpMultiplyAdd &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_58_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Wmma_3_01Shape___00_01cutlass_1_1int4b__t_00_01LayoutA___00_01cutlass_16fd808a90b3cf9d7cfc99f30888ca3fe.html" target="_self">Wmma&lt; Shape_, cutlass::int4b_t, LayoutA_, cutlass::int4b_t, LayoutB_, int32_t, LayoutC_, cutlass::arch::OpMultiplyAdd &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_59_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Wmma_3_01Shape___00_01cutlass_1_1uint1b__t_00_01LayoutA___00_01cutlass_c80a7ea4d219cd9b13b560b493338028.html" target="_self">Wmma&lt; Shape_, cutlass::uint1b_t, LayoutA_, cutlass::uint1b_t, LayoutB_, int32_t, LayoutC_, cutlass::arch::OpXorPopc &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_60_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Wmma_3_01Shape___00_01int8__t_00_01LayoutA___00_01int8__t_00_01LayoutB_505c57bb6818a941dc16f00cf35a9ec0.html" target="_self">Wmma&lt; Shape_, int8_t, LayoutA_, int8_t, LayoutB_, int32_t, LayoutC_, cutlass::arch::OpMultiplyAdd &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_0_61_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1arch_1_1Wmma_3_01Shape___00_01uint8__t_00_01LayoutA___00_01uint8__t_00_01Layout219a464a1248ebfc37aa29bcb10cb1b0.html" target="_self">Wmma&lt; Shape_, uint8_t, LayoutA_, uint8_t, LayoutB_, int32_t, LayoutC_, cutlass::arch::OpMultiplyAdd &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_1_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_1_" class="arrow" onclick="toggleFolder('0_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1device__memory.html" target="_self">device_memory</a></td><td class="desc"></td></tr>
+<tr id="row_0_1_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_1_0_" class="arrow" onclick="toggleFolder('0_1_0_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1device__memory_1_1allocation.html" target="_self">allocation</a></td><td class="desc">Device allocation abstraction that tracks size and capacity </td></tr>
+<tr id="row_0_1_0_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1device__memory_1_1allocation_1_1deleter.html" target="_self">deleter</a></td><td class="desc">Delete functor for CUDA device memory </td></tr>
+<tr id="row_0_2_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_2_" class="arrow" onclick="toggleFolder('0_2_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1epilogue.html" target="_self">epilogue</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_2_0_" class="arrow" onclick="toggleFolder('0_2_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1epilogue_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_0_" class="arrow" onclick="toggleFolder('0_2_0_0_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1Convert.html" target="_self">Convert</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1Convert_1_1Params.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_0_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_1_" class="arrow" onclick="toggleFolder('0_2_0_1_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1LinearCombination.html" target="_self">LinearCombination</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_1_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1LinearCombination_1_1Params.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_0_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_2_" class="arrow" onclick="toggleFolder('0_2_0_2_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1LinearCombinationClamp.html" target="_self">LinearCombinationClamp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_2_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1LinearCombinationClamp_1_1Params.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_0_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_3_" class="arrow" onclick="toggleFolder('0_2_0_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1LinearCombinationRelu.html" target="_self">LinearCombinationRelu</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_3_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1LinearCombinationRelu_1_1Params.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_0_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_4_" class="arrow" onclick="toggleFolder('0_2_0_4_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1LinearCombinationRelu_3_01ElementOutput___00_01Count_00_01int_00_01float_00_01Round_01_4.html" target="_self">LinearCombinationRelu&lt; ElementOutput_, Count, int, float, Round &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_4_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1LinearCombinationRelu_3_01ElementOutput___00_01Count_00_00274a94522c46cd041d0b10d484e2ef3.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_0_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_0_5_" class="arrow" onclick="toggleFolder('0_2_0_5_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1thread_1_1ReductionOpPlus.html" target="_self">ReductionOpPlus</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_0_5_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1thread_1_1ReductionOpPlus_1_1Params.html" target="_self">Params</a></td><td class="desc">Host-constructable parameters structure </td></tr>
+<tr id="row_0_2_1_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_2_1_" class="arrow" onclick="toggleFolder('0_2_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1epilogue_1_1threadblock.html" target="_self">threadblock</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_0_" class="arrow" onclick="toggleFolder('0_2_1_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1epilogue_1_1threadblock_1_1detail.html" target="_self">detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement.html" target="_self">RowArrangement</a></td><td class="desc"><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement.html" title="RowArrangement determines how one or more warps cover a region of consecutive rows. ">RowArrangement</a> determines how one or more warps cover a region of consecutive rows </td></tr>
+<tr id="row_0_2_1_0_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement_3_01Shape_00_01WarpsRemaini91159e6f7e123d881e3ec45101fa4f81.html" target="_self">RowArrangement&lt; Shape, WarpsRemaining, ElementsPerAccess, ElementSize, false &gt;</a></td><td class="desc"><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement.html" title="RowArrangement determines how one or more warps cover a region of consecutive rows. ">RowArrangement</a> in which each warp's access is a 1D tiled arrangement </td></tr>
+<tr id="row_0_2_1_0_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_2_1_0_2_" class="arrow" onclick="toggleFolder('0_2_1_0_2_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement_3_01Shape_00_01WarpsRemaini6d8790249bf12cac580da73bb37eb791.html" target="_self">RowArrangement&lt; Shape, WarpsRemaining, ElementsPerAccess, ElementSize, true &gt;</a></td><td class="desc"><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement.html" title="RowArrangement determines how one or more warps cover a region of consecutive rows. ">RowArrangement</a> in which each warp's access is a 2D tiled arrangement </td></tr>
+<tr id="row_0_2_1_0_2_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1detail_1_1RowArrangement_3_01Shape_00_01WarpsRemainief28e98b3f284469f271d28aba73de2e.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultEpilogueComplexTensorOp.html" target="_self">DefaultEpilogueComplexTensorOp</a></td><td class="desc">Defines sensible defaults for epilogues for TensorOps </td></tr>
+<tr id="row_0_2_1_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultEpilogueSimt.html" target="_self">DefaultEpilogueSimt</a></td><td class="desc">Defines sensible defaults for epilogues for SimtOps </td></tr>
+<tr id="row_0_2_1_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultEpilogueTensorOp.html" target="_self">DefaultEpilogueTensorOp</a></td><td class="desc">Defines sensible defaults for epilogues for TensorOps </td></tr>
+<tr id="row_0_2_1_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultEpilogueVoltaTensorOp.html" target="_self">DefaultEpilogueVoltaTensorOp</a></td><td class="desc">Defines sensible defaults for epilogues for TensorOps </td></tr>
+<tr id="row_0_2_1_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultEpilogueWmmaTensorOp.html" target="_self">DefaultEpilogueWmmaTensorOp</a></td><td class="desc">Defines sensible defaults for epilogues for WMMA TensorOps </td></tr>
+<tr id="row_0_2_1_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultInterleavedEpilogueTensorOp.html" target="_self">DefaultInterleavedEpilogueTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_7_" class="arrow" onclick="toggleFolder('0_2_1_7_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultInterleavedThreadMapTensorOp.html" target="_self">DefaultInterleavedThreadMapTensorOp</a></td><td class="desc">Defines the optimal thread map for TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_7_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultInterleavedThreadMapTensorOp_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_8_" class="arrow" onclick="toggleFolder('0_2_1_8_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapSimt.html" target="_self">DefaultThreadMapSimt</a></td><td class="desc">Defines the optimal thread map for SIMT accumulator layouts </td></tr>
+<tr id="row_0_2_1_8_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapSimt_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_9_" class="arrow" onclick="toggleFolder('0_2_1_9_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapTensorOp.html" target="_self">DefaultThreadMapTensorOp</a></td><td class="desc">Defines the optimal thread map for TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_9_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapTensorOp_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapVoltaTensorOp.html" target="_self">DefaultThreadMapVoltaTensorOp</a></td><td class="desc">Defines the optimal thread map for TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_11_" class="arrow" onclick="toggleFolder('0_2_1_11_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapVoltaTensorOp_3_01ThreadblockShape__95db04b7b72e34283958bd7fbf851d16.html" target="_self">DefaultThreadMapVoltaTensorOp&lt; ThreadblockShape_, WarpShape_, PartitionsK, ElementOutput_, ElementsPerAccess, float &gt;</a></td><td class="desc">Defines the optimal thread map for TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_11_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapVoltaTensorOp_3_01ThreadblockShape__52116c60c62f0fd520071558e42b814f.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_12_" class="arrow" onclick="toggleFolder('0_2_1_12_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapVoltaTensorOp_3_01ThreadblockShape__d58c94abc36b7c5c109b55202c6992e7.html" target="_self">DefaultThreadMapVoltaTensorOp&lt; ThreadblockShape_, WarpShape_, PartitionsK, ElementOutput_, ElementsPerAccess, half_t &gt;</a></td><td class="desc">Defines the optimal thread map for TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_12_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapVoltaTensorOp_3_01ThreadblockShape__4433cc988100e98097a748d2670fb0fc.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_13_" class="arrow" onclick="toggleFolder('0_2_1_13_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapWmmaTensorOp.html" target="_self">DefaultThreadMapWmmaTensorOp</a></td><td class="desc">Defines the optimal thread map for Wmma TensorOp accumulator layouts </td></tr>
+<tr id="row_0_2_1_13_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DefaultThreadMapWmmaTensorOp_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_14_" class="arrow" onclick="toggleFolder('0_2_1_14_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1DirectEpilogueTensorOp.html" target="_self">DirectEpilogueTensorOp</a></td><td class="desc"><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1Epilogue.html" title="Epilogue operator without splitk. ">Epilogue</a> operator </td></tr>
+<tr id="row_0_2_1_14_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DirectEpilogueTensorOp_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure for host-constructible state </td></tr>
+<tr id="row_0_2_1_14_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1DirectEpilogueTensorOp_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared storage allocation needed by the epilogue </td></tr>
+<tr id="row_0_2_1_15_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1Epilogue.html" target="_self">Epilogue</a></td><td class="desc"><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1Epilogue.html" title="Epilogue operator without splitk. ">Epilogue</a> operator without splitk </td></tr>
+<tr id="row_0_2_1_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_16_" class="arrow" onclick="toggleFolder('0_2_1_16_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1EpilogueBase.html" target="_self">EpilogueBase</a></td><td class="desc">Base class for epilogues defining warp-level </td></tr>
+<tr id="row_0_2_1_16_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1EpilogueBase_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared storage allocation needed by the epilogue </td></tr>
+<tr id="row_0_2_1_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_17_" class="arrow" onclick="toggleFolder('0_2_1_17_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1InterleavedEpilogue.html" target="_self">InterleavedEpilogue</a></td><td class="desc"><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1Epilogue.html" title="Epilogue operator without splitk. ">Epilogue</a> operator without splitk </td></tr>
+<tr id="row_0_2_1_17_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedEpilogue_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared storage allocation needed by the epilogue </td></tr>
+<tr id="row_0_2_1_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_18_" class="arrow" onclick="toggleFolder('0_2_1_18_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedOutputTileThreadMap.html" target="_self">InterleavedOutputTileThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_18_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedOutputTileThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_19_" class="arrow" onclick="toggleFolder('0_2_1_19_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1InterleavedPredicatedTileIterator.html" target="_self">InterleavedPredicatedTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_19_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedPredicatedTileIterator_1_1Mask.html" target="_self">Mask</a></td><td class="desc"><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedPredicatedTileIterator_1_1Mask.html" title="Mask object. ">Mask</a> object </td></tr>
+<tr id="row_0_2_1_19_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1InterleavedPredicatedTileIterator_1_1Params.html" target="_self">Params</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_20_" class="arrow" onclick="toggleFolder('0_2_1_20_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1OutputTileOptimalThreadMap.html" target="_self">OutputTileOptimalThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_20_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1OutputTileOptimalThreadMap_1_1CompactedThreadMap.html" target="_self">CompactedThreadMap</a></td><td class="desc">Compacted thread map in which the 4D region is contiguous </td></tr>
+<tr id="row_0_2_1_20_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1OutputTileOptimalThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_21_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1OutputTileShape.html" target="_self">OutputTileShape</a></td><td class="desc">Tuple defining point in output tile </td></tr>
+<tr id="row_0_2_1_22_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1OutputTileThreadMap.html" target="_self">OutputTileThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_1_23_" class="arrow" onclick="toggleFolder('0_2_1_23_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1PredicatedTileIterator.html" target="_self">PredicatedTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_23_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1PredicatedTileIterator_1_1Mask.html" target="_self">Mask</a></td><td class="desc"><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1PredicatedTileIterator_1_1Mask.html" title="Mask object. ">Mask</a> object </td></tr>
+<tr id="row_0_2_1_23_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1threadblock_1_1PredicatedTileIterator_1_1Params.html" target="_self">Params</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_1_24_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1threadblock_1_1SharedLoadIterator.html" target="_self">SharedLoadIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_2_2_" class="arrow" onclick="toggleFolder('0_2_2_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1epilogue_1_1warp.html" target="_self">warp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorComplexTensorOp.html" target="_self">FragmentIteratorComplexTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorComplexTensorOp_3_01WarpShape___00_01Operato8cf03c624cf3210c71b7cbd580b080f8.html" target="_self">FragmentIteratorComplexTensorOp&lt; WarpShape_, OperatorShape_, OperatorElementC_, OperatorFragmentC_, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorSimt.html" target="_self">FragmentIteratorSimt</a></td><td class="desc">Fragment iterator for SIMT accumulator arrangements </td></tr>
+<tr id="row_0_2_2_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorSimt_3_01WarpShape___00_01Operator___00_01la3f2abc523201c1b0228df99119ab88e1.html" target="_self">FragmentIteratorSimt&lt; WarpShape_, Operator_, layout::RowMajor, MmaSimtPolicy_ &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorTensorOp.html" target="_self">FragmentIteratorTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorTensorOp_3_01WarpShape___00_01OperatorShape_e459aab140a2ce78336e584f95886726.html" target="_self">FragmentIteratorTensorOp&lt; WarpShape_, OperatorShape_, OperatorElementC_, OperatorFragmentC_, layout::ColumnMajorInterleaved&lt; InterleavedK &gt; &gt;</a></td><td class="desc">Dedicated to interleaved layout </td></tr>
+<tr id="row_0_2_2_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorTensorOp_3_01WarpShape___00_01OperatorShape_5e78dabe303f20d76b00c600aab61eda.html" target="_self">FragmentIteratorTensorOp&lt; WarpShape_, OperatorShape_, OperatorElementC_, OperatorFragmentC_, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorVoltaTensorOp.html" target="_self">FragmentIteratorVoltaTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1Gdb805a2dc5571ac3b66e0fe6ffdcede2.html" target="_self">FragmentIteratorVoltaTensorOp&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, float, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1G16e08718cffa0989cce3fe8dbc4b075b.html" target="_self">FragmentIteratorVoltaTensorOp&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, half_t, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorWmmaTensorOp.html" target="_self">FragmentIteratorWmmaTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_11_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1FragmentIteratorWmmaTensorOp_3_01WarpShape___00_01OperatorShfdb1f120c6797383663f9fd11d0fc599.html" target="_self">FragmentIteratorWmmaTensorOp&lt; WarpShape_, OperatorShape_, OperatorElementC_, OperatorFragmentC_, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major shared memory </td></tr>
+<tr id="row_0_2_2_12_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1SimtPolicy.html" target="_self">SimtPolicy</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_13_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1SimtPolicy_3_01WarpShape___00_01Operator___00_01layout_1_1Rcef1c60e23e997017ae176c92931151d.html" target="_self">SimtPolicy&lt; WarpShape_, Operator_, layout::RowMajor, MmaSimtPolicy_ &gt;</a></td><td class="desc">Partial specialization for row-major </td></tr>
+<tr id="row_0_2_2_14_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TensorOpPolicy.html" target="_self">TensorOpPolicy</a></td><td class="desc">Policy details related to the epilogue </td></tr>
+<tr id="row_0_2_2_15_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TensorOpPolicy_3_01WarpShape_00_01OperatorShape_00_01layout69549d10c3610d943987eb90e827bc05.html" target="_self">TensorOpPolicy&lt; WarpShape, OperatorShape, layout::ColumnMajorInterleaved&lt; InterleavedK &gt; &gt;</a></td><td class="desc">Partial specialization for column-major-interleaved </td></tr>
+<tr id="row_0_2_2_16_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TensorOpPolicy_3_01WarpShape_00_01OperatorShape_00_01layout_1_1RowMajor_01_4.html" target="_self">TensorOpPolicy&lt; WarpShape, OperatorShape, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major </td></tr>
+<tr id="row_0_2_2_17_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorSimt.html" target="_self">TileIteratorSimt</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_18_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorSimt_3_01WarpShape___00_01Operator___00_01Elemenf2bd262ed3e202b25d5802d83965bf3b.html" target="_self">TileIteratorSimt&lt; WarpShape_, Operator_, Element_, layout::RowMajor, MmaSimtPolicy_ &gt;</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_19_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorTensorOp.html" target="_self">TileIteratorTensorOp</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_2_20_" class="arrow" onclick="toggleFolder('0_2_2_20_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorTensorOp_3_01WarpShape___00_01OperatorShape___003cbb32beb84b4984cb7853662096d289.html" target="_self">TileIteratorTensorOp&lt; WarpShape_, OperatorShape_, Element_, layout::RowMajor &gt;</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_20_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TileIteratorTensorOp_3_01WarpShape___00_01OperatorShape___05f11e023c9e6ee5f7a888fa4c5bbf6d1.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_21_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TileIteratorVoltaTensorOp.html" target="_self">TileIteratorVoltaTensorOp</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_22_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_2_22_" class="arrow" onclick="toggleFolder('0_2_2_22_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1GemmS2fe0c60b727c738c622c18fc3dd76644.html" target="_self">TileIteratorVoltaTensorOp&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, float, layout::RowMajor &gt;</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_22_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TileIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1Gemm770cbca45441d295d5d7433e8222a700.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_2_2_23_" class="arrow" onclick="toggleFolder('0_2_2_23_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1GemmSa0ceeeddc22575876eb977da7f5416a8.html" target="_self">TileIteratorVoltaTensorOp&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, half_t, layout::RowMajor &gt;</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_23_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1TileIteratorVoltaTensorOp_3_01WarpShape___00_01gemm_1_1Gemmffcab2297c8de8d0013602a39c525b78.html" target="_self">Detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_2_24_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorWmmaTensorOp.html" target="_self">TileIteratorWmmaTensorOp</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_25_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1warp_1_1TileIteratorWmmaTensorOp_3_01WarpShape___00_01OperatorShape_fd6a91cd8bbd07ecd1344326b830e3a4.html" target="_self">TileIteratorWmmaTensorOp&lt; WarpShape_, OperatorShape_, OperatorFragment_, layout::RowMajor &gt;</a></td><td class="desc">Template for reading and writing tiles of accumulators to shared memory </td></tr>
+<tr id="row_0_2_2_26_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1VoltaTensorOpPolicy.html" target="_self">VoltaTensorOpPolicy</a></td><td class="desc">Policy details related to the epilogue </td></tr>
+<tr id="row_0_2_2_27_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1VoltaTensorOpPolicy_3_01WarpShape___00_01gemm_1_1GemmShape_136ce744d4c1c6e8707f5a9785196194.html" target="_self">VoltaTensorOpPolicy&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, float, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major </td></tr>
+<tr id="row_0_2_2_28_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1warp_1_1VoltaTensorOpPolicy_3_01WarpShape___00_01gemm_1_1GemmShape_1d48185f49e4d066f8e9327bf0856b7f.html" target="_self">VoltaTensorOpPolicy&lt; WarpShape_, gemm::GemmShape&lt; 32, 32, 4 &gt;, half_t, layout::RowMajor &gt;</a></td><td class="desc">Partial specialization for row-major </td></tr>
+<tr id="row_0_2_3_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_2_3_" class="arrow" onclick="toggleFolder('0_2_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1epilogue_1_1EpilogueWorkspace.html" target="_self">EpilogueWorkspace</a></td><td class="desc"></td></tr>
+<tr id="row_0_2_3_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1EpilogueWorkspace_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_2_3_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1epilogue_1_1EpilogueWorkspace_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared storage allocation needed by the epilogue </td></tr>
+<tr id="row_0_3_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_3_" class="arrow" onclick="toggleFolder('0_3_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm.html" target="_self">gemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_3_0_" class="arrow" onclick="toggleFolder('0_3_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1device.html" target="_self">device</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration.html" target="_self">DefaultGemmConfiguration</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassSimt_00_01ArchTag286687c5e6abe22d241f789fe344a465.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassSimt, ArchTag, ElementA, ElementB, ElementC, ElementAccumulator &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassSimt_00_01ArchTag3026e48abb8c905d1cc6d13d669700e4.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassSimt, ArchTag, int8_t, int8_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arc567cad318a31d04b70ea615d6321decd.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm70, ElementA, ElementB, ElementC, ElementAccumulator &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arcde61af9be1337dac1fdb210e7e7a6e01.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, ElementA, ElementB, ElementC, ElementAccumulator &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arc485a4f0b5a7d2d4ab2c1a24da6328048.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, int4b_t, int4b_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arc8e2604a56dff3a7595da9ee0604ae55e.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, int4b_t, uint4b_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arc4fada4957d463c80a2831e47f28157c4.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, int8_t, int8_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arc8ab5fd2693c6a6ec43e447acb07f784c.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, int8_t, uint8_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arcffcf31256aed23d4d8d0eab627bc0cad.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, uint4b_t, int4b_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arcb2e258b7bd321c633dd65d3ebcf6414a.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, uint4b_t, uint4b_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_11_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arcb27bf218007928652d5b803193eab473.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, uint8_t, int8_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_12_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassTensorOp_00_01arcfea0f3503156e8e3fba6456f0cedafdd.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassTensorOp, arch::Sm75, uint8_t, uint8_t, ElementC, int32_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_13_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1DefaultGemmConfiguration_3_01arch_1_1OpClassWmmaTensorOp_00_0884059ecad03bea3e86c4cf722226097.html" target="_self">DefaultGemmConfiguration&lt; arch::OpClassWmmaTensorOp, ArchTag, ElementA, ElementB, ElementC, ElementAccumulator &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_14_" class="arrow" onclick="toggleFolder('0_3_0_14_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1Gemm.html" target="_self">Gemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_14_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1Gemm_1_1Arguments.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_15_" class="arrow" onclick="toggleFolder('0_3_0_15_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1Gemm_3_01ElementA___00_01LayoutA___00_01ElementB___00_01Layout4d0960ae6b1d1bf19e6239dbd002249c.html" target="_self">Gemm&lt; ElementA_, LayoutA_, ElementB_, LayoutB_, ElementC_, layout::ColumnMajor, ElementAccumulator_, OperatorClass_, ArchTag_, ThreadblockShape_, WarpShape_, InstructionShape_, EpilogueOutputOp_, ThreadblockSwizzle_, Stages, AlignmentA, AlignmentB, SplitKSerial, Operator_, IsBetaZero &gt;</a></td><td class="desc">Partial specialization for column-major output exchanges problem size and operand </td></tr>
+<tr id="row_0_3_0_15_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1Gemm_3_01ElementA___00_01LayoutA___00_01ElementB___00_01Layou1b211cc9c97c022d8fe10f2dd32c8709.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_16_" class="arrow" onclick="toggleFolder('0_3_0_16_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmBatched.html" target="_self">GemmBatched</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_16_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmBatched_1_1Arguments.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_17_" class="arrow" onclick="toggleFolder('0_3_0_17_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmBatched_3_01ElementA___00_01LayoutA___00_01ElementB___00_0c9bb6f4463ab6085e6008b5d5ad6abfd.html" target="_self">GemmBatched&lt; ElementA_, LayoutA_, ElementB_, LayoutB_, ElementC_, layout::ColumnMajor, ElementAccumulator_, OperatorClass_, ArchTag_, ThreadblockShape_, WarpShape_, InstructionShape_, EpilogueOutputOp_, ThreadblockSwizzle_, Stages, AlignmentA, AlignmentB, Operator_ &gt;</a></td><td class="desc">Partial specialization for column-major output exchanges problem size and operand </td></tr>
+<tr id="row_0_3_0_17_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmBatched_3_01ElementA___00_01LayoutA___00_01ElementB___00_213d78696663f4231cd52c6a277c60e5.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_18_" class="arrow" onclick="toggleFolder('0_3_0_18_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmComplex.html" target="_self">GemmComplex</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_18_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmComplex_1_1Arguments.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_19_" class="arrow" onclick="toggleFolder('0_3_0_19_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmComplex_3_01ElementA___00_01LayoutA___00_01ElementB___00_07c56401b4df75709ae636675d9980a9a.html" target="_self">GemmComplex&lt; ElementA_, LayoutA_, ElementB_, LayoutB_, ElementC_, layout::ColumnMajor, ElementAccumulator_, OperatorClass_, ArchTag_, ThreadblockShape_, WarpShape_, InstructionShape_, EpilogueOutputOp_, ThreadblockSwizzle_, Stages, TransformA, TransformB, SplitKSerial &gt;</a></td><td class="desc">Partial specialization for column-major output exchanges problem size and operand </td></tr>
+<tr id="row_0_3_0_19_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmComplex_3_01ElementA___00_01LayoutA___00_01ElementB___00_a3923967cafb5cb9774c320dc24baa77.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_20_" class="arrow" onclick="toggleFolder('0_3_0_20_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmSplitKParallel.html" target="_self">GemmSplitKParallel</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_0_20_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmSplitKParallel_1_1Arguments.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_0_21_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_0_21_" class="arrow" onclick="toggleFolder('0_3_0_21_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1device_1_1GemmSplitKParallel_3_01ElementA___00_01LayoutA___00_01ElementBbe7c1f7154ad5b5bf9d4d28301e2b457.html" target="_self">GemmSplitKParallel&lt; ElementA_, LayoutA_, ElementB_, LayoutB_, ElementC_, layout::ColumnMajor, ElementAccumulator_, OperatorClass_, ArchTag_, ThreadblockShape_, WarpShape_, InstructionShape_, EpilogueOutputOp_, ConvertScaledOp_, ReductionOp_, ThreadblockSwizzle_, Stages, kAlignmentA, kAlignmentB, Operator_ &gt;</a></td><td class="desc">Partial specialization for column-major output </td></tr>
+<tr id="row_0_3_0_21_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1device_1_1GemmSplitKParallel_3_01ElementA___00_01LayoutA___00_01Elementafcb1aeaf2035a7ac769d7acc233423b.html" target="_self">Arguments</a></td><td class="desc">Argument structure </td></tr>
+<tr id="row_0_3_1_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_3_1_" class="arrow" onclick="toggleFolder('0_3_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1kernel.html" target="_self">kernel</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_1_0_" class="arrow" onclick="toggleFolder('0_3_1_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1kernel_1_1detail.html" target="_self">detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1detail_1_1GemvBatchedStridedEpilogueScaling.html" target="_self">GemvBatchedStridedEpilogueScaling</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm.html" target="_self">DefaultGemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm_3_01ElementA_00_01layout_1_1ColumnMajorInterleave661fe54d13cc2c9153dcdf31e4beaa30.html" target="_self">DefaultGemm&lt; ElementA, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, kAlignmentA, ElementB, layout::RowMajorInterleaved&lt; InterleavedK &gt;, kAlignmentB, ElementC, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, int32_t, arch::OpClassTensorOp, arch::Sm75, ThreadblockShape, WarpShape, InstructionShape, EpilogueOutputOp, ThreadblockSwizzle, 2, SplitKSerial, Operator, IsBetaZero &gt;</a></td><td class="desc">Partial specialization for Turing Integer Matrix Multiply Interleaved layout </td></tr>
+<tr id="row_0_3_1_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm_3_01ElementA_00_01LayoutA_00_01kAlignmentA_00_01Edd80343e6570718ed237122e4ebf7fb5.html" target="_self">DefaultGemm&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementC, layout::RowMajor, ElementAccumulator, arch::OpClassSimt, ArchTag, ThreadblockShape, WarpShape, GemmShape&lt; 1, 1, 1 &gt;, EpilogueOutputOp, ThreadblockSwizzle, 2, SplitKSerial, Operator &gt;</a></td><td class="desc">Partial specialization for SIMT </td></tr>
+<tr id="row_0_3_1_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm_3_01ElementA_00_01LayoutA_00_01kAlignmentA_00_01E044b039b2fe402f29b04a9f5feee5342.html" target="_self">DefaultGemm&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementC, layout::RowMajor, ElementAccumulator, arch::OpClassTensorOp, arch::Sm70, ThreadblockShape, WarpShape, GemmShape&lt; 8, 8, 4 &gt;, EpilogueOutputOp, ThreadblockSwizzle, 2, SplitKSerial, Operator &gt;</a></td><td class="desc">Partial specialization for Volta architecture </td></tr>
+<tr id="row_0_3_1_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm_3_01ElementA_00_01LayoutA_00_01kAlignmentA_00_01E5d78d37a9ae2ec08d7d477d571df036e.html" target="_self">DefaultGemm&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementC, layout::RowMajor, ElementAccumulator, arch::OpClassTensorOp, arch::Sm75, ThreadblockShape, WarpShape, InstructionShape, EpilogueOutputOp, ThreadblockSwizzle, 2, SplitKSerial, Operator &gt;</a></td><td class="desc">Partial specialization for Turing Architecture </td></tr>
+<tr id="row_0_3_1_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemm_3_01int8__t_00_01LayoutA_00_01kAlignmentA_00_01inf48440732c1c5f42ddbfaba179861815.html" target="_self">DefaultGemm&lt; int8_t, LayoutA, kAlignmentA, int8_t, LayoutB, kAlignmentB, ElementC, LayoutC, ElementAccumulator, arch::OpClassSimt, ArchTag, ThreadblockShape, WarpShape, GemmShape&lt; 1, 1, 4 &gt;, EpilogueOutputOp, ThreadblockSwizzle, 2, SplitKSerial, Operator, false &gt;</a></td><td class="desc">Partial specialization for SIMT DP4A </td></tr>
+<tr id="row_0_3_1_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemmSplitKParallel.html" target="_self">DefaultGemmSplitKParallel</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1DefaultGemv.html" target="_self">DefaultGemv</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_1_9_" class="arrow" onclick="toggleFolder('0_3_1_9_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1Gemm.html" target="_self">Gemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_9_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1Gemm_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_3_1_9_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="unioncutlass_1_1gemm_1_1kernel_1_1Gemm_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared memory storage structure </td></tr>
+<tr id="row_0_3_1_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_1_10_" class="arrow" onclick="toggleFolder('0_3_1_10_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1GemmBatched.html" target="_self">GemmBatched</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_10_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1GemmBatched_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_3_1_10_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="unioncutlass_1_1gemm_1_1kernel_1_1GemmBatched_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared memory storage structure </td></tr>
+<tr id="row_0_3_1_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_1_11_" class="arrow" onclick="toggleFolder('0_3_1_11_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1GemmSplitKParallel.html" target="_self">GemmSplitKParallel</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_1_11_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1kernel_1_1GemmSplitKParallel_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_3_1_11_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="unioncutlass_1_1gemm_1_1kernel_1_1GemmSplitKParallel_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared memory storage structure </td></tr>
+<tr id="row_0_3_2_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_3_2_" class="arrow" onclick="toggleFolder('0_3_2_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_2_0_" class="arrow" onclick="toggleFolder('0_3_2_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1thread_1_1detail.html" target="_self">detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1EnableMma__Crow__SM60.html" target="_self">EnableMma_Crow_SM60</a></td><td class="desc">Determines whether to enable thread::Gemm&lt;&gt; specializations compatible with SM50 </td></tr>
+<tr id="row_0_3_2_0_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2.html" target="_self">Mma_HFMA2</a></td><td class="desc">Structure to compute the matrix product for HFMA </td></tr>
+<tr id="row_0_3_2_0_2_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1ColumnMajor_00_72621f7ab9ae4a4ba4fe9725cf8e89c1.html" target="_self">Mma_HFMA2&lt; Shape, layout::ColumnMajor, layout::ColumnMajor, layout::ColumnMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_3_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1ColumnMajor_00_94c813e3bbfb6f9857c155166f772687.html" target="_self">Mma_HFMA2&lt; Shape, layout::ColumnMajor, layout::ColumnMajor, layout::RowMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_4_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1ColumnMajor_00_17070298bc4cced0a1b98aee2bb6b455.html" target="_self">Mma_HFMA2&lt; Shape, layout::ColumnMajor, layout::RowMajor, layout::ColumnMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_5_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1ColumnMajor_00_bf6d29bb09a025e7b96942809743e28a.html" target="_self">Mma_HFMA2&lt; Shape, layout::ColumnMajor, layout::RowMajor, layout::RowMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_6_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1RowMajor_00_01l26a133b13650c1d058273e3649f60f04.html" target="_self">Mma_HFMA2&lt; Shape, layout::RowMajor, layout::ColumnMajor, layout::ColumnMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_7_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1RowMajor_00_01lbba3a796be96a0276693ef6b259ecc4a.html" target="_self">Mma_HFMA2&lt; Shape, layout::RowMajor, layout::ColumnMajor, layout::RowMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_8_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1RowMajor_00_01l2aa4d2fd2e940e0d0cf7c47bc8f6017c.html" target="_self">Mma_HFMA2&lt; Shape, layout::RowMajor, layout::RowMajor, layout::ColumnMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_9_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01layout_1_1RowMajor_00_01l086c058a15d6c79558e4f3d9ff1dc148.html" target="_self">Mma_HFMA2&lt; Shape, layout::RowMajor, layout::RowMajor, layout::RowMajor, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_10_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01LayoutA_00_01LayoutB_00_0e1104c65871c539155bd3a0c7631928b.html" target="_self">Mma_HFMA2&lt; Shape, LayoutA, LayoutB, layout::ColumnMajor, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_0_11_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1detail_1_1Mma__HFMA2_3_01Shape_00_01LayoutA_00_01LayoutB_00_07ac147cb320ee0d28ff8e78eb4cd330e.html" target="_self">Mma_HFMA2&lt; Shape, LayoutA, LayoutB, layout::RowMajor, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_2_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma.html" target="_self">Mma</a></td><td class="desc">Structure to compute the matrix product </td></tr>
+<tr id="row_0_3_2_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma_3_01Shape___00_01ElementA___00_01LayoutA___00_01ElementB_e41c1cd6078b6d1347fac239b0639d56.html" target="_self">Mma&lt; Shape_, ElementA_, LayoutA_, ElementB_, LayoutB_, ElementC_, LayoutC_, arch::OpMultiplyAdd, bool &gt;</a></td><td class="desc">Gemplate that handles conventional layouts for FFMA and DFMA GEMM </td></tr>
+<tr id="row_0_3_2_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma_3_01Shape___00_01half__t_00_01LayoutA_00_01half__t_00_01L066c9d2371712cdf0cac099ca9bcc578.html" target="_self">Mma&lt; Shape_, half_t, LayoutA, half_t, LayoutB, half_t, LayoutC, arch::OpMultiplyAdd &gt;</a></td><td class="desc">Structure to compute the matrix product </td></tr>
+<tr id="row_0_3_2_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma_3_01Shape___00_01half__t_00_01LayoutA___00_01half__t_00_088f0e99e501b6012297eb30b4e89bcea.html" target="_self">Mma&lt; Shape_, half_t, LayoutA_, half_t, LayoutB_, half_t, layout::RowMajor, arch::OpMultiplyAdd, typename platform::enable_if&lt; detail::EnableMma_Crow_SM60&lt; LayoutA_, LayoutB_ &gt;::value &gt;::type &gt;</a></td><td class="desc">Computes matrix product when C is row-major </td></tr>
+<tr id="row_0_3_2_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma_3_01Shape___00_01int8__t_00_01layout_1_1ColumnMajor_00_013f3785e722edc6e9aab6f866309b8623.html" target="_self">Mma&lt; Shape_, int8_t, layout::ColumnMajor, int8_t, layout::RowMajor, int32_t, LayoutC_, arch::OpMultiplyAdd, int8_t &gt;</a></td><td class="desc">Gemplate that handles conventional layouts for IDP4A </td></tr>
+<tr id="row_0_3_2_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1Mma_3_01Shape___00_01int8__t_00_01layout_1_1RowMajor_00_01int89c659e7faf47264972bdba6cd80f42b.html" target="_self">Mma&lt; Shape_, int8_t, layout::RowMajor, int8_t, layout::ColumnMajor, int32_t, LayoutC_, arch::OpMultiplyAdd, bool &gt;</a></td><td class="desc">Gemplate that handles conventional layouts for IDP4A </td></tr>
+<tr id="row_0_3_2_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1thread_1_1MmaGeneric.html" target="_self">MmaGeneric</a></td><td class="desc">Gemplate that handles all packed matrix layouts </td></tr>
+<tr id="row_0_3_3_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_3_3_" class="arrow" onclick="toggleFolder('0_3_3_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1threadblock.html" target="_self">threadblock</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultGemvCore.html" target="_self">DefaultGemvCore</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMma.html" target="_self">DefaultMma</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMma_3_01ElementA_00_01LayoutA_00_01kAlignmentA_0010764e1fd5a3251a57eddafbd83eab8e.html" target="_self">DefaultMma&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementAccumulator, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, OperatorClass, ArchTag, ThreadblockShape, WarpShape, InstructionShape, 2, Operator, true &gt;</a></td><td class="desc">Specialization for column-major-interleaved output </td></tr>
+<tr id="row_0_3_3_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMma_3_01ElementA_00_01LayoutA_00_01kAlignmentA_00c67c16f9881e4f2fda76d8ed83ebabd6.html" target="_self">DefaultMma&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementAccumulator, layout::RowMajor, arch::OpClassSimt, ArchTag, ThreadblockShape, WarpShape, InstructionShape, 2, Operator, false &gt;</a></td><td class="desc">Specialization for row-major output (OperatorClass Simt) </td></tr>
+<tr id="row_0_3_3_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMma_3_01ElementA_00_01LayoutA_00_01kAlignmentA_00ce36642cae579bce6605ff8edde3c6ab.html" target="_self">DefaultMma&lt; ElementA, LayoutA, kAlignmentA, ElementB, LayoutB, kAlignmentB, ElementAccumulator, layout::RowMajor, arch::OpClassTensorOp, ArchTag, ThreadblockShape, WarpShape, InstructionShape, 2, Operator, false &gt;</a></td><td class="desc">Specialization for row-major output (OperatorClass Simt) </td></tr>
+<tr id="row_0_3_3_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMma_3_01int8__t_00_01LayoutA_00_01kAlignmentA_00_07e7230d4011ada5e22cfcb29103b696.html" target="_self">DefaultMma&lt; int8_t, LayoutA, kAlignmentA, int8_t, LayoutB, kAlignmentB, ElementAccumulator, layout::RowMajor, arch::OpClassSimt, ArchTag, ThreadblockShape, WarpShape, GemmShape&lt; 1, 1, 4 &gt;, 2, Operator, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore.html" target="_self">DefaultMmaCore</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmShab94a11a77dd0565102710907089acee0.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 1 &gt;, ElementA_, layout::ColumnMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmShafafd5c61db86cbfe90863578ddd11092.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 1 &gt;, ElementA_, layout::ColumnMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha46446d1e3871e31d2e728f710d78c8c1.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 1 &gt;, ElementA_, layout::ColumnMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_, &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha8da7a0cfbbe859b701fdd9f2b8566aa7.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 1 &gt;, ElementA_, layout::RowMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_11_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha84e9f8afb6a4ca9f5dcd219b182d16e7.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 1 &gt;, ElementA_, layout::RowMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_12_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha2c0d0b7cdb5c4bcb11e83c058eb65345.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 4 &gt;, int8_t, layout::ColumnMajor, int8_t, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc">Partial specialization: </td></tr>
+<tr id="row_0_3_3_13_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha34a52cc7b2942e8c290f0032b6779b52.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 4 &gt;, int8_t, layout::ColumnMajor, int8_t, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_14_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmShaaf312aafe9da92ea9d417bcc12a8e7dc.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 4 &gt;, int8_t, layout::RowMajor, int8_t, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc">Partial specialization: </td></tr>
+<tr id="row_0_3_3_15_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha863d4139ccaa713bc4bde32c425f4067.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 1, 1, 4 &gt;, int8_t, layout::RowMajor, int8_t, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassSimt, 2, Operator_ &gt;</a></td><td class="desc">Partial specialization: </td></tr>
+<tr id="row_0_3_3_16_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmShaf03a122202ad10acdc96f280106d678b.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 8, 8, 4 &gt;, ElementA_, layout::ColumnMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_17_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha69bef08ea63dd930f99d9788105873dd.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 8, 8, 4 &gt;, ElementA_, layout::ColumnMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_18_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmSha3adf608332a8c9ee7014fced0da8a9ca.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 8, 8, 4 &gt;, ElementA_, layout::RowMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_19_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01GemmShab7edfba3cdf43a07e3c4d719d87565a4.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, GemmShape&lt; 8, 8, 4 &gt;, ElementA_, layout::RowMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_20_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01Instruc803d38bc1e4618c07c47f54c87ae2678.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, InstructionShape_, ElementA_, layout::ColumnMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_21_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01Instrucf60fe02fcdd80d28b7fd419133465dcc.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, InstructionShape_, ElementA_, layout::ColumnMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_22_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01Instruc2bf00737f4ad0a9da9a8be6d3e66c152.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, InstructionShape_, ElementA_, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, ElementB_, layout::RowMajorInterleaved&lt; InterleavedK &gt;, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_, AccumulatorsInRowMajor &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_23_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01Instruc24092ddc01fc83dabb7db4c14880fe60.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, InstructionShape_, ElementA_, layout::RowMajor, ElementB_, layout::ColumnMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_24_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1DefaultMmaCore_3_01Shape___00_01WarpShape___00_01Instruc4fee9f2965b8468bfb42b94a74527d22.html" target="_self">DefaultMmaCore&lt; Shape_, WarpShape_, InstructionShape_, ElementA_, layout::RowMajor, ElementB_, layout::RowMajor, ElementC_, LayoutC_, arch::OpClassTensorOp, 2, Operator_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_25_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemmBatchedIdentityThreadblockSwizzle.html" target="_self">GemmBatchedIdentityThreadblockSwizzle</a></td><td class="desc">Threadblock swizzling function for batched GEMMs </td></tr>
+<tr id="row_0_3_3_26_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemmHorizontalThreadblockSwizzle.html" target="_self">GemmHorizontalThreadblockSwizzle</a></td><td class="desc">Threadblock swizzling function for GEMMs </td></tr>
+<tr id="row_0_3_3_27_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemmIdentityThreadblockSwizzle.html" target="_self">GemmIdentityThreadblockSwizzle</a></td><td class="desc">Threadblock swizzling function for GEMMs </td></tr>
+<tr id="row_0_3_3_28_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemmSplitKHorizontalThreadblockSwizzle.html" target="_self">GemmSplitKHorizontalThreadblockSwizzle</a></td><td class="desc">Threadblock swizzling function for split-K GEMMs </td></tr>
+<tr id="row_0_3_3_29_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemmSplitKIdentityThreadblockSwizzle.html" target="_self">GemmSplitKIdentityThreadblockSwizzle</a></td><td class="desc">Threadblock swizzling function for split-K GEMMs </td></tr>
+<tr id="row_0_3_3_30_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1threadblock_1_1Gemv.html" target="_self">Gemv</a></td><td class="desc">Structure to compute the matrix-vector product using SIMT math instructions </td></tr>
+<tr id="row_0_3_3_31_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1GemvBatchedStridedThreadblockDefaultSwizzle.html" target="_self">GemvBatchedStridedThreadblockDefaultSwizzle</a></td><td class="desc">Threadblock swizzling function for batched GEMVs </td></tr>
+<tr id="row_0_3_3_32_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_3_32_" class="arrow" onclick="toggleFolder('0_3_3_32_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1threadblock_1_1MmaBase.html" target="_self">MmaBase</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_3_32_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1threadblock_1_1MmaBase_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc">Shared storage object needed by threadblock-scoped GEMM </td></tr>
+<tr id="row_0_3_3_33_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1threadblock_1_1MmaPipelined.html" target="_self">MmaPipelined</a></td><td class="desc">Structure to compute the matrix product targeting CUDA cores and SIMT math instructions </td></tr>
+<tr id="row_0_3_3_34_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1threadblock_1_1MmaPolicy.html" target="_self">MmaPolicy</a></td><td class="desc">Policy object describing MmaTensorOp </td></tr>
+<tr id="row_0_3_3_35_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1threadblock_1_1MmaSingleStage.html" target="_self">MmaSingleStage</a></td><td class="desc">Structure to compute the matrix product targeting CUDA cores and SIMT math instructions </td></tr>
+<tr id="row_0_3_4_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_3_4_" class="arrow" onclick="toggleFolder('0_3_4_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1gemm_1_1warp.html" target="_self">warp</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1DefaultMmaTensorOp.html" target="_self">DefaultMmaTensorOp</a></td><td class="desc">Partial specialization for m-by-n-by-kgroup </td></tr>
+<tr id="row_0_3_4_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaComplexTensorOp.html" target="_self">MmaComplexTensorOp</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaComplexTensorOp_3_01Shape___00_01complex_3_01RealElementA_01_146441010dad1f40eb51b6dae3ded216.html" target="_self">MmaComplexTensorOp&lt; Shape_, complex&lt; RealElementA &gt;, LayoutA_, complex&lt; RealElementB &gt;, LayoutB_, complex&lt; RealElementC &gt;, LayoutC_, Policy_, TransformA, TransformB, Enable &gt;</a></td><td class="desc">Partial specialization for complex*complex+complex =&gt; complex using real-valued TensorOps </td></tr>
+<tr id="row_0_3_4_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimt.html" target="_self">MmaSimt</a></td><td class="desc">Structure to compute the matrix product targeting CUDA cores and SIMT math instructions </td></tr>
+<tr id="row_0_3_4_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaSimtPolicy.html" target="_self">MmaSimtPolicy</a></td><td class="desc">Describes the arrangement and configuration of per-lane operations in warp-level matrix multiply </td></tr>
+<tr id="row_0_3_4_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator.html" target="_self">MmaSimtTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kA_00_01Element_67ca7e11a38e38f2c51b84767654a90f.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kA, Element_, layout::ColumnMajor, Policy_, PartitionsK, PartitionGroupSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kA_00_01Element_f0ce904a9294556f15e1cc9cf7c99a93.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kA, Element_, layout::ColumnMajorInterleaved&lt; 4 &gt;, Policy_, PartitionsK, PartitionGroupSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kB_00_01Element_ea0a4e7ce3cd5d25cabf79383efdf4d9.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kB, Element_, layout::RowMajor, Policy_, PartitionsK, PartitionGroupSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kB_00_01Element_ada156b62fcbdce47009c5bf1321c92c.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kB, Element_, layout::RowMajorInterleaved&lt; 4 &gt;, Policy_, PartitionsK, PartitionGroupSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kC_00_01Element_4ccafbc821b3a55cd532602442a74031.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kC, Element_, layout::ColumnMajor, Policy_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_11_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaSimtTileIterator_3_01Shape___00_01Operand_1_1kC_00_01Element_8f92ea79e85febb67169c4b2d94b1b20.html" target="_self">MmaSimtTileIterator&lt; Shape_, Operand::kC, Element_, layout::RowMajor, Policy_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_12_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOp.html" target="_self">MmaTensorOp</a></td><td class="desc">Structure to compute the matrix product targeting CUDA cores and SIMT math instructions </td></tr>
+<tr id="row_0_3_4_13_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator.html" target="_self">MmaTensorOpAccumulatorTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_14_" class="arrow" onclick="toggleFolder('0_3_4_14_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___008f607b871a2b3d854eb4def64712c042.html" target="_self">MmaTensorOpAccumulatorTileIterator&lt; Shape_, Element_, cutlass::layout::ColumnMajor, InstructionShape_, OpDelta_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_14_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___0d35fa5dc4e4b4f72784c943fd857fc1d.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_15_" class="arrow" onclick="toggleFolder('0_3_4_15_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___00027dabdc144edd6276f664ca74088510.html" target="_self">MmaTensorOpAccumulatorTileIterator&lt; Shape_, Element_, cutlass::layout::ColumnMajorInterleaved&lt; InterleavedN &gt;, InstructionShape_, OpDelta_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_15_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___03822d9be37f3725022005a5434441f22.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_16_" class="arrow" onclick="toggleFolder('0_3_4_16_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___006c39f57875e0aa9d0ad82c8043ed8b98.html" target="_self">MmaTensorOpAccumulatorTileIterator&lt; Shape_, Element_, cutlass::layout::RowMajor, InstructionShape_, OpDelta_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_16_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpAccumulatorTileIterator_3_01Shape___00_01Element___093b5d2838ac5a742704ef62b5c8688f0.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_17_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator.html" target="_self">MmaTensorOpMultiplicandTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_18_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0b84f53cd44b339eccc12067c9f86e11c.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::ColumnMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_19_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0e52ad425e1ee3e68544873f66733237b.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::ColumnMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_20_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___039819fb3ccd43786d556c2c9669508ef.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::RowMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_21_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0352e0dcab42bc8360606874e00173556.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::RowMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_22_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_22_" class="arrow" onclick="toggleFolder('0_3_4_22_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0ed7daaeba1c095e77f68533d4d2c475c.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::TensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, 64 &gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_22_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___07638f8b7761f6e2e2e6918e2c05e739.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_23_" class="arrow" onclick="toggleFolder('0_3_4_23_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0c7d419c589d601ce4eb603be566fea21.html" target="_self">MmaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::TensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, InstructionShape_, OpDelta_, 32, PartitionsK_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_23_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand___0784c74bd670999ec23ad8ef9dc55777.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_24_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaTensorOpPolicy.html" target="_self">MmaTensorOpPolicy</a></td><td class="desc">Policy </td></tr>
+<tr id="row_0_3_4_25_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOp.html" target="_self">MmaVoltaTensorOp</a></td><td class="desc">Structure to compute the matrix product targeting CUDA cores and SIMT math instructions </td></tr>
+<tr id="row_0_3_4_26_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_26_" class="arrow" onclick="toggleFolder('0_3_4_26_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpAccumulatorTileIterator.html" target="_self">MmaVoltaTensorOpAccumulatorTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_26_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpAccumulatorTileIterator_1_1Policy.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_27_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_28_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operan0d3248553e52cd61ed8a2b3b12a20343.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand::kA, Element_, cutlass::layout::ColumnMajorVoltaTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_29_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_29_" class="arrow" onclick="toggleFolder('0_3_4_29_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operan34be8e21a40af3ebd2dc3dff460dca72.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand::kA, Element_, cutlass::layout::VoltaTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_29_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Opera33cdf53848564e894d4407637dc86caf.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_30_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operand734577b7e54a074d143aba59828c2f2.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand::kB, Element_, cutlass::layout::RowMajorVoltaTensorOpMultiplicandBCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_31_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_31_" class="arrow" onclick="toggleFolder('0_3_4_31_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operan16c56cdc2dda5eeb996af8ec0242d501.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand::kB, Element_, cutlass::layout::VoltaTensorOpMultiplicandBCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_31_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Opera6fa6d2d3725bb3ec613d5c527ea3ffe7.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_32_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operan5a221944f4a0e16ccab77ba684856942.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::ColumnMajorVoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, KBlock &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_33_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operandcc9821c435540895138bc9af495f321.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::RowMajorVoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, KBlock &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_34_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_3_4_34_" class="arrow" onclick="toggleFolder('0_3_4_34_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operana2f40b28f0d2286b84d86f7238d67b52.html" target="_self">MmaVoltaTensorOpMultiplicandTileIterator&lt; Shape_, Operand_, Element_, cutlass::layout::VoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, KBlock &gt;, InstructionShape_, OpDelta_, 32 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_4_34_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1MmaVoltaTensorOpMultiplicandTileIterator_3_01Shape___00_01Operafa294175b280756dd8388f9ffe7b72c4.html" target="_self">Policy</a></td><td class="desc">Internal structure of iterator - made public to enable introspection </td></tr>
+<tr id="row_0_3_4_35_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1warp_1_1WarpSize.html" target="_self">WarpSize</a></td><td class="desc">Query the number of threads per warp </td></tr>
+<tr id="row_0_3_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1BatchedGemmCoord.html" target="_self">BatchedGemmCoord</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1GemmCoord.html" target="_self">GemmCoord</a></td><td class="desc"></td></tr>
+<tr id="row_0_3_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1gemm_1_1GemmShape.html" target="_self">GemmShape</a></td><td class="desc">Shape of a matrix multiply-add operation </td></tr>
+<tr id="row_0_4_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_4_" class="arrow" onclick="toggleFolder('0_4_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1layout.html" target="_self">layout</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1ColumnMajor.html" target="_self">ColumnMajor</a></td><td class="desc">Mapping function for column-major matrices </td></tr>
+<tr id="row_0_4_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorBlockLinear.html" target="_self">ColumnMajorBlockLinear</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorInterleaved.html" target="_self">ColumnMajorInterleaved</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorTensorOpMultiplicandCongruous.html" target="_self">ColumnMajorTensorOpMultiplicandCongruous</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorTensorOpMultiplicandCrosswise.html" target="_self">ColumnMajorTensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorVoltaTensorOpMultiplicandBCongruous.html" target="_self">ColumnMajorVoltaTensorOpMultiplicandBCongruous</a></td><td class="desc">Template mapping a column-major view of pitch-linear memory to <a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCongruous.html" title="Template based on element size (in bits) - defined in terms of pitch-linear memory. ">VoltaTensorOpMultiplicandCongruous</a> </td></tr>
+<tr id="row_0_4_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorVoltaTensorOpMultiplicandCongruous.html" target="_self">ColumnMajorVoltaTensorOpMultiplicandCongruous</a></td><td class="desc">Template mapping a column-major view of pitch-linear memory to <a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCongruous.html" title="Template based on element size (in bits) - defined in terms of pitch-linear memory. ">VoltaTensorOpMultiplicandCongruous</a> </td></tr>
+<tr id="row_0_4_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ColumnMajorVoltaTensorOpMultiplicandCrosswise.html" target="_self">ColumnMajorVoltaTensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1ContiguousMatrix.html" target="_self">ContiguousMatrix</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1GeneralMatrix.html" target="_self">GeneralMatrix</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1LayoutTranspose.html" target="_self">LayoutTranspose</a></td><td class="desc">Defines transposes of matrix layouts </td></tr>
+<tr id="row_0_4_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1LayoutTranspose_3_01layout_1_1ColumnMajor_01_4.html" target="_self">LayoutTranspose&lt; layout::ColumnMajor &gt;</a></td><td class="desc">Transpose of column-major is row-major </td></tr>
+<tr id="row_0_4_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1LayoutTranspose_3_01layout_1_1RowMajor_01_4.html" target="_self">LayoutTranspose&lt; layout::RowMajor &gt;</a></td><td class="desc">Transpose of row-major is column-major </td></tr>
+<tr id="row_0_4_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1PackedVectorLayout.html" target="_self">PackedVectorLayout</a></td><td class="desc">Tensor layout for densely packed vectors </td></tr>
+<tr id="row_0_4_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1PitchLinear.html" target="_self">PitchLinear</a></td><td class="desc">Mapping function for pitch-linear memory </td></tr>
+<tr id="row_0_4_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1PitchLinearCoord.html" target="_self">PitchLinearCoord</a></td><td class="desc">Coordinate in pitch-linear space </td></tr>
+<tr id="row_0_4_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1PitchLinearShape.html" target="_self">PitchLinearShape</a></td><td class="desc">Template defining a shape used by pitch-linear operators </td></tr>
+<tr id="row_0_4_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1RowMajor.html" target="_self">RowMajor</a></td><td class="desc">Mapping function for row-major matrices </td></tr>
+<tr id="row_0_4_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorBlockLinear.html" target="_self">RowMajorBlockLinear</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorInterleaved.html" target="_self">RowMajorInterleaved</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorTensorOpMultiplicandCongruous.html" target="_self">RowMajorTensorOpMultiplicandCongruous</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_21_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorTensorOpMultiplicandCrosswise.html" target="_self">RowMajorTensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_22_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorVoltaTensorOpMultiplicandBCongruous.html" target="_self">RowMajorVoltaTensorOpMultiplicandBCongruous</a></td><td class="desc">Template mapping a row-major view of pitch-linear memory to <a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCongruous.html" title="Template based on element size (in bits) - defined in terms of pitch-linear memory. ">VoltaTensorOpMultiplicandCongruous</a> </td></tr>
+<tr id="row_0_4_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorVoltaTensorOpMultiplicandCongruous.html" target="_self">RowMajorVoltaTensorOpMultiplicandCongruous</a></td><td class="desc">Template mapping a row-major view of pitch-linear memory to <a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCongruous.html" title="Template based on element size (in bits) - defined in terms of pitch-linear memory. ">VoltaTensorOpMultiplicandCongruous</a> </td></tr>
+<tr id="row_0_4_24_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1RowMajorVoltaTensorOpMultiplicandCrosswise.html" target="_self">RowMajorVoltaTensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_25_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1TensorCxRSKx.html" target="_self">TensorCxRSKx</a></td><td class="desc">Mapping function for 4-D CxRSKx tensors </td></tr>
+<tr id="row_0_4_26_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1TensorNCHW.html" target="_self">TensorNCHW</a></td><td class="desc">Mapping function for 4-D NCHW tensors </td></tr>
+<tr id="row_0_4_27_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1TensorNCxHWx.html" target="_self">TensorNCxHWx</a></td><td class="desc">Mapping function for 4-D NC/xHWx tensors </td></tr>
+<tr id="row_0_4_28_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1layout_1_1TensorNHWC.html" target="_self">TensorNHWC</a></td><td class="desc">Mapping function for 4-D NHWC tensors </td></tr>
+<tr id="row_0_4_29_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicand.html" target="_self">TensorOpMultiplicand</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_30_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicandColumnMajorInterleaved.html" target="_self">TensorOpMultiplicandColumnMajorInterleaved</a></td><td class="desc">Template based on element size (in bits) - defined in terms of pitch-linear memory </td></tr>
+<tr id="row_0_4_31_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicandCongruous.html" target="_self">TensorOpMultiplicandCongruous</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_32_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicandCongruous_3_0132_00_01Crosswise_01_4.html" target="_self">TensorOpMultiplicandCongruous&lt; 32, Crosswise &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_33_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicandCrosswise.html" target="_self">TensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_4_34_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1TensorOpMultiplicandRowMajorInterleaved.html" target="_self">TensorOpMultiplicandRowMajorInterleaved</a></td><td class="desc">Template based on element size (in bits) - defined in terms of pitch-linear memory </td></tr>
+<tr id="row_0_4_35_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandBCongruous.html" target="_self">VoltaTensorOpMultiplicandBCongruous</a></td><td class="desc">Template based on element size (in bits) - defined in terms of pitch-linear memory </td></tr>
+<tr id="row_0_4_36_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCongruous.html" target="_self">VoltaTensorOpMultiplicandCongruous</a></td><td class="desc">Template based on element size (in bits) - defined in terms of pitch-linear memory </td></tr>
+<tr id="row_0_4_37_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1layout_1_1VoltaTensorOpMultiplicandCrosswise.html" target="_self">VoltaTensorOpMultiplicandCrosswise</a></td><td class="desc"></td></tr>
+<tr id="row_0_5_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_5_" class="arrow" onclick="toggleFolder('0_5_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1library.html" target="_self">library</a></td><td class="desc"></td></tr>
+<tr id="row_0_5_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmArguments.html" target="_self">GemmArguments</a></td><td class="desc">Arguments for GEMM </td></tr>
+<tr id="row_0_5_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmArrayArguments.html" target="_self">GemmArrayArguments</a></td><td class="desc">Arguments for GEMM - used by all the GEMM operations </td></tr>
+<tr id="row_0_5_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmArrayConfiguration.html" target="_self">GemmArrayConfiguration</a></td><td class="desc">Configuration for batched GEMM in which multiple matrix products are computed </td></tr>
+<tr id="row_0_5_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmBatchedConfiguration.html" target="_self">GemmBatchedConfiguration</a></td><td class="desc">Configuration for batched GEMM in which multiple matrix products are computed </td></tr>
+<tr id="row_0_5_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmConfiguration.html" target="_self">GemmConfiguration</a></td><td class="desc">Configuration for basic GEMM operations </td></tr>
+<tr id="row_0_5_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmDescription.html" target="_self">GemmDescription</a></td><td class="desc">Description of all GEMM computations </td></tr>
+<tr id="row_0_5_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmPlanarComplexBatchedConfiguration.html" target="_self">GemmPlanarComplexBatchedConfiguration</a></td><td class="desc">Batched complex valued GEMM in which real and imaginary parts are separated by a stride </td></tr>
+<tr id="row_0_5_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1GemmPlanarComplexConfiguration.html" target="_self">GemmPlanarComplexConfiguration</a></td><td class="desc">Complex valued GEMM in which real and imaginary parts are separated by a stride </td></tr>
+<tr id="row_0_5_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1library_1_1Manifest.html" target="_self">Manifest</a></td><td class="desc"><a class="el" href="classcutlass_1_1library_1_1Manifest.html" title="Manifest of CUTLASS Library. ">Manifest</a> of CUTLASS Library </td></tr>
+<tr id="row_0_5_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1MathInstructionDescription.html" target="_self">MathInstructionDescription</a></td><td class="desc"></td></tr>
+<tr id="row_0_5_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1library_1_1Operation.html" target="_self">Operation</a></td><td class="desc">Base class for all device-wide operations </td></tr>
+<tr id="row_0_5_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1OperationDescription.html" target="_self">OperationDescription</a></td><td class="desc">High-level description of an operation </td></tr>
+<tr id="row_0_5_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1TensorDescription.html" target="_self">TensorDescription</a></td><td class="desc">Structure describing the properties of a tensor </td></tr>
+<tr id="row_0_5_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1library_1_1TileDescription.html" target="_self">TileDescription</a></td><td class="desc">Structure describing the tiled structure of a GEMM-like computation </td></tr>
+<tr id="row_0_6_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_6_" class="arrow" onclick="toggleFolder('0_6_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1platform.html" target="_self">platform</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1aligned__chunk.html" target="_self">aligned_chunk</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1aligned__storage.html" target="_self">aligned_storage</a></td><td class="desc">Std::aligned_storage </td></tr>
+<tr id="row_0_6_2_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_6_2_" class="arrow" onclick="toggleFolder('0_6_2_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of.html" target="_self">alignment_of</a></td><td class="desc">Std::alignment_of </td></tr>
+<tr id="row_0_6_2_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_1_1pad.html" target="_self">pad</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01const_01value__t_01_4.html" target="_self">alignment_of&lt; const value_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01const_01volatile_01value__t_01_4.html" target="_self">alignment_of&lt; const volatile value_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01double2_01_4.html" target="_self">alignment_of&lt; double2 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01double4_01_4.html" target="_self">alignment_of&lt; double4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01float4_01_4.html" target="_self">alignment_of&lt; float4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01int4_01_4.html" target="_self">alignment_of&lt; int4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01long4_01_4.html" target="_self">alignment_of&lt; long4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01longlong2_01_4.html" target="_self">alignment_of&lt; longlong2 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01longlong4_01_4.html" target="_self">alignment_of&lt; longlong4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01uint4_01_4.html" target="_self">alignment_of&lt; uint4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01ulong4_01_4.html" target="_self">alignment_of&lt; ulong4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01ulonglong2_01_4.html" target="_self">alignment_of&lt; ulonglong2 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01ulonglong4_01_4.html" target="_self">alignment_of&lt; ulonglong4 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1alignment__of_3_01volatile_01value__t_01_4.html" target="_self">alignment_of&lt; volatile value_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1bool__constant.html" target="_self">bool_constant</a></td><td class="desc">Std::bool_constant </td></tr>
+<tr id="row_0_6_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1conditional.html" target="_self">conditional</a></td><td class="desc">Std::conditional (true specialization) </td></tr>
+<tr id="row_0_6_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1conditional_3_01false_00_01T_00_01F_01_4.html" target="_self">conditional&lt; false, T, F &gt;</a></td><td class="desc">Std::conditional (false specialization) </td></tr>
+<tr id="row_0_6_20_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1default__delete.html" target="_self">default_delete</a></td><td class="desc">Default deleter </td></tr>
+<tr id="row_0_6_21_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1default__delete_3_01T[]_4.html" target="_self">default_delete&lt; T[]&gt;</a></td><td class="desc">Partial specialization for deleting array types </td></tr>
+<tr id="row_0_6_22_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1enable__if.html" target="_self">enable_if</a></td><td class="desc">Std::enable_if (true specialization) </td></tr>
+<tr id="row_0_6_23_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1enable__if_3_01false_00_01T_01_4.html" target="_self">enable_if&lt; false, T &gt;</a></td><td class="desc">Std::enable_if (false specialization) </td></tr>
+<tr id="row_0_6_24_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1integral__constant.html" target="_self">integral_constant</a></td><td class="desc">Std::integral_constant </td></tr>
+<tr id="row_0_6_25_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__arithmetic.html" target="_self">is_arithmetic</a></td><td class="desc">Std::is_arithmetic </td></tr>
+<tr id="row_0_6_26_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__base__of.html" target="_self">is_base_of</a></td><td class="desc">Std::is_base_of </td></tr>
+<tr id="row_0_6_27_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_6_27_" class="arrow" onclick="toggleFolder('0_6_27_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__base__of__helper.html" target="_self">is_base_of_helper</a></td><td class="desc">Helper for std::is_base_of </td></tr>
+<tr id="row_0_6_27_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__base__of__helper_1_1dummy.html" target="_self">dummy</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_28_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__floating__point.html" target="_self">is_floating_point</a></td><td class="desc">Std::is_floating_point </td></tr>
+<tr id="row_0_6_29_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__fundamental.html" target="_self">is_fundamental</a></td><td class="desc">Std::is_fundamental </td></tr>
+<tr id="row_0_6_30_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral.html" target="_self">is_integral</a></td><td class="desc">Std::is_integral </td></tr>
+<tr id="row_0_6_31_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01char_01_4.html" target="_self">is_integral&lt; char &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_32_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01const_01T_01_4.html" target="_self">is_integral&lt; const T &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_33_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01const_01volatile_01T_01_4.html" target="_self">is_integral&lt; const volatile T &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_34_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01int_01_4.html" target="_self">is_integral&lt; int &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_35_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01long_01_4.html" target="_self">is_integral&lt; long &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_36_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01long_01long_01_4.html" target="_self">is_integral&lt; long long &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_37_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01short_01_4.html" target="_self">is_integral&lt; short &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_38_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01signed_01char_01_4.html" target="_self">is_integral&lt; signed char &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_39_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01unsigned_01char_01_4.html" target="_self">is_integral&lt; unsigned char &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_40_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01unsigned_01int_01_4.html" target="_self">is_integral&lt; unsigned int &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_41_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01unsigned_01long_01_4.html" target="_self">is_integral&lt; unsigned long &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_42_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01unsigned_01long_01long_01_4.html" target="_self">is_integral&lt; unsigned long long &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_43_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01unsigned_01short_01_4.html" target="_self">is_integral&lt; unsigned short &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_44_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__integral_3_01volatile_01T_01_4.html" target="_self">is_integral&lt; volatile T &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_45_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__pointer.html" target="_self">is_pointer</a></td><td class="desc">Std::is_pointer </td></tr>
+<tr id="row_0_6_46_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__pointer__helper.html" target="_self">is_pointer_helper</a></td><td class="desc">Helper for std::is_pointer (false specialization) </td></tr>
+<tr id="row_0_6_47_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__pointer__helper_3_01T_01_5_01_4.html" target="_self">is_pointer_helper&lt; T * &gt;</a></td><td class="desc">Helper for std::is_pointer (true specialization) </td></tr>
+<tr id="row_0_6_48_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__same.html" target="_self">is_same</a></td><td class="desc">Std::is_same (false specialization) </td></tr>
+<tr id="row_0_6_49_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__same_3_01A_00_01A_01_4.html" target="_self">is_same&lt; A, A &gt;</a></td><td class="desc">Std::is_same (true specialization) </td></tr>
+<tr id="row_0_6_50_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__trivially__copyable.html" target="_self">is_trivially_copyable</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_51_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__void.html" target="_self">is_void</a></td><td class="desc">Std::is_void </td></tr>
+<tr id="row_0_6_52_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__volatile.html" target="_self">is_volatile</a></td><td class="desc">Std::is_volatile </td></tr>
+<tr id="row_0_6_53_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1is__volatile_3_01volatile_01T_01_4.html" target="_self">is_volatile&lt; volatile T &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_6_54_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1nullptr__t.html" target="_self">nullptr_t</a></td><td class="desc">Std::nullptr_t </td></tr>
+<tr id="row_0_6_55_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1remove__const.html" target="_self">remove_const</a></td><td class="desc">Std::remove_const (non-const specialization) </td></tr>
+<tr id="row_0_6_56_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1remove__const_3_01const_01T_01_4.html" target="_self">remove_const&lt; const T &gt;</a></td><td class="desc">Std::remove_const (const specialization) </td></tr>
+<tr id="row_0_6_57_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1remove__cv.html" target="_self">remove_cv</a></td><td class="desc">Std::remove_cv </td></tr>
+<tr id="row_0_6_58_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1remove__volatile.html" target="_self">remove_volatile</a></td><td class="desc">Std::remove_volatile (non-volatile specialization) </td></tr>
+<tr id="row_0_6_59_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1platform_1_1remove__volatile_3_01volatile_01T_01_4.html" target="_self">remove_volatile&lt; volatile T &gt;</a></td><td class="desc">Std::remove_volatile (volatile specialization) </td></tr>
+<tr id="row_0_6_60_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1platform_1_1unique__ptr.html" target="_self">unique_ptr</a></td><td class="desc">Std::unique_ptr </td></tr>
+<tr id="row_0_7_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_7_" class="arrow" onclick="toggleFolder('0_7_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reduction.html" target="_self">reduction</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_7_0_" class="arrow" onclick="toggleFolder('0_7_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reduction_1_1kernel.html" target="_self">kernel</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_0_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_7_0_0_" class="arrow" onclick="toggleFolder('0_7_0_0_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1reduction_1_1kernel_1_1ReduceSplitK.html" target="_self">ReduceSplitK</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_0_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1kernel_1_1ReduceSplitK_1_1Params.html" target="_self">Params</a></td><td class="desc"><a class="el" href="structcutlass_1_1reduction_1_1kernel_1_1ReduceSplitK_1_1Params.html" title="Params structure. ">Params</a> structure </td></tr>
+<tr id="row_0_7_0_0_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1kernel_1_1ReduceSplitK_1_1SharedStorage.html" target="_self">SharedStorage</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_1_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_7_1_" class="arrow" onclick="toggleFolder('0_7_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reduction_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_1_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce.html" target="_self">Reduce</a></td><td class="desc">Structure to compute the thread level reduction </td></tr>
+<tr id="row_0_7_1_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce_3_01plus_3_01half__t_01_4_00_01AlignedArray_3_01half__t_00_01N_01_4_01_4.html" target="_self">Reduce&lt; plus&lt; half_t &gt;, AlignedArray&lt; half_t, N &gt; &gt;</a></td><td class="desc">Partial specializations of <a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce.html" title="Structure to compute the thread level reduction. ">Reduce</a> for AlignedArray&lt;half_t, N&gt; </td></tr>
+<tr id="row_0_7_1_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce_3_01plus_3_01half__t_01_4_00_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">Reduce&lt; plus&lt; half_t &gt;, Array&lt; half_t, N &gt; &gt;</a></td><td class="desc">Partial specializations of <a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce.html" title="Structure to compute the thread level reduction. ">Reduce</a> for Array&lt;half_t, N&gt; </td></tr>
+<tr id="row_0_7_1_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce_3_01plus_3_01T_01_4_00_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">Reduce&lt; plus&lt; T &gt;, Array&lt; T, N &gt; &gt;</a></td><td class="desc">Partial specialization of <a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce.html" title="Structure to compute the thread level reduction. ">Reduce</a> for Array&lt;T, N&gt; </td></tr>
+<tr id="row_0_7_1_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce_3_01plus_3_01T_01_4_00_01T_01_4.html" target="_self">Reduce&lt; plus&lt; T &gt;, T &gt;</a></td><td class="desc">Partial Specialization of <a class="el" href="structcutlass_1_1reduction_1_1thread_1_1Reduce.html" title="Structure to compute the thread level reduction. ">Reduce</a> for "plus" (a functional operator) </td></tr>
+<tr id="row_0_7_1_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_7_1_5_" class="arrow" onclick="toggleFolder('0_7_1_5_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1ReduceAdd.html" target="_self">ReduceAdd</a></td><td class="desc">Mixed-precision reduction </td></tr>
+<tr id="row_0_7_1_5_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1thread_1_1ReduceAdd_1_1Params.html" target="_self">Params</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1BatchedReduction.html" target="_self">BatchedReduction</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_3_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_7_3_" class="arrow" onclick="toggleFolder('0_7_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1BatchedReductionTraits.html" target="_self">BatchedReductionTraits</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_3_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1BatchedReductionTraits_1_1Params.html" target="_self">Params</a></td><td class="desc"></td></tr>
+<tr id="row_0_7_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reduction_1_1DefaultBlockSwizzle.html" target="_self">DefaultBlockSwizzle</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_8_" class="arrow" onclick="toggleFolder('0_8_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference.html" target="_self">reference</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_8_0_" class="arrow" onclick="toggleFolder('0_8_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1detail.html" target="_self">detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_0_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1detail_1_1Cast.html" target="_self">Cast</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_0_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1detail_1_1Cast_3_01float_00_01int8__t_01_4.html" target="_self">Cast&lt; float, int8_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_0_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1detail_1_1Cast_3_01float_00_01uint8__t_01_4.html" target="_self">Cast&lt; float, uint8_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_8_1_" class="arrow" onclick="toggleFolder('0_8_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1device.html" target="_self">device</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_" class="arrow" onclick="toggleFolder('0_8_1_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1device_1_1detail.html" target="_self">detail</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_0_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_0_" class="arrow" onclick="toggleFolder('0_8_1_0_0_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1RandomGaussianFunc.html" target="_self">RandomGaussianFunc</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_0_0_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1RandomGaussianFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_1_" class="arrow" onclick="toggleFolder('0_8_1_0_1_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1RandomUniformFunc.html" target="_self">RandomUniformFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_1_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1RandomUniformFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_2_" class="arrow" onclick="toggleFolder('0_8_1_0_2_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorCopyDiagonalInFunc.html" target="_self">TensorCopyDiagonalInFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_2_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorCopyDiagonalInFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_3_" class="arrow" onclick="toggleFolder('0_8_1_0_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorCopyDiagonalOutFunc.html" target="_self">TensorCopyDiagonalOutFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_3_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorCopyDiagonalOutFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_4_" class="arrow" onclick="toggleFolder('0_8_1_0_4_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillDiagonalFunc.html" target="_self">TensorFillDiagonalFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_4_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillDiagonalFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_5_" class="arrow" onclick="toggleFolder('0_8_1_0_5_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillLinearFunc.html" target="_self">TensorFillLinearFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_5_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillLinearFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_6_" class="arrow" onclick="toggleFolder('0_8_1_0_6_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillRandomGaussianFunc.html" target="_self">TensorFillRandomGaussianFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_6_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillRandomGaussianFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_7_" class="arrow" onclick="toggleFolder('0_8_1_0_7_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillRandomUniformFunc.html" target="_self">TensorFillRandomUniformFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_7_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorFillRandomUniformFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_8_" class="arrow" onclick="toggleFolder('0_8_1_0_8_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorUpdateDiagonalFunc.html" target="_self">TensorUpdateDiagonalFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_8_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorUpdateDiagonalFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_0_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_0_9_" class="arrow" onclick="toggleFolder('0_8_1_0_9_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorUpdateOffDiagonalFunc.html" target="_self">TensorUpdateOffDiagonalFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_1_0_9_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1detail_1_1TensorUpdateOffDiagonalFunc_1_1Params.html" target="_self">Params</a></td><td class="desc">Parameters structure </td></tr>
+<tr id="row_0_8_1_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_8_1_1_" class="arrow" onclick="toggleFolder('0_8_1_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1device_1_1kernel.html" target="_self">kernel</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_1_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span id="arr_0_8_1_1_0_" class="arrow" onclick="toggleFolder('0_8_1_1_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1device_1_1kernel_1_1detail.html" target="_self">detail</a></td><td class="desc">Defines several helpers </td></tr>
+<tr id="row_0_8_1_1_0_0_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1kernel_1_1detail_1_1TensorForEachHelper.html" target="_self">TensorForEachHelper</a></td><td class="desc">Helper to perform for-each operation </td></tr>
+<tr id="row_0_8_1_1_0_1_" style="display:none;"><td class="entry"><span style="width:96px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1kernel_1_1detail_1_1TensorForEachHelper_3_01Func_00_01Rank_00_010_01_4.html" target="_self">TensorForEachHelper&lt; Func, Rank, 0 &gt;</a></td><td class="desc">Helper to perform for-each operation </td></tr>
+<tr id="row_0_8_1_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_8_1_2_" class="arrow" onclick="toggleFolder('0_8_1_2_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1device_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_2_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1thread_1_1Gemm.html" target="_self">Gemm</a></td><td class="desc">Thread-level blocked general matrix product </td></tr>
+<tr id="row_0_8_1_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1BlockForEach.html" target="_self">BlockForEach</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1Gemm.html" target="_self">Gemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_1_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01Layout4e016ab7cfc644acd7cb4ae770339773.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, AccumulatorType, arch::OpMultiplyAdd &gt;</a></td><td class="desc">Partial specialization for multiply-add </td></tr>
+<tr id="row_0_8_1_6_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01Layout30b72addd464a2ca4a26785cbfd77a8e.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, AccumulatorType, arch::OpMultiplyAddSaturate &gt;</a></td><td class="desc">Partial specialization for multiply-add-saturate </td></tr>
+<tr id="row_0_8_1_7_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01Layout660562b232f408218828ca5915b7e73a.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, AccumulatorType, arch::OpXorPopc &gt;</a></td><td class="desc">Partial specialization for XOR-popc </td></tr>
+<tr id="row_0_8_1_8_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1TensorDiagonalForEach.html" target="_self">TensorDiagonalForEach</a></td><td class="desc">Launches a kernel calling a functor for each element along a tensor's diagonal </td></tr>
+<tr id="row_0_8_1_9_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1device_1_1TensorForEach.html" target="_self">TensorForEach</a></td><td class="desc">Launches a kernel calling a functor for each element in a tensor's index space </td></tr>
+<tr id="row_0_8_2_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_8_2_" class="arrow" onclick="toggleFolder('0_8_2_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1host.html" target="_self">host</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_2_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_8_2_0_" class="arrow" onclick="toggleFolder('0_8_2_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1reference_1_1host_1_1detail.html" target="_self">detail</a></td><td class="desc">Defines several helpers </td></tr>
+<tr id="row_0_8_2_0_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1RandomGaussianFunc.html" target="_self">RandomGaussianFunc</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_2_0_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1RandomGaussianFunc_3_01complex_3_01Element_01_4_01_4.html" target="_self">RandomGaussianFunc&lt; complex&lt; Element &gt; &gt;</a></td><td class="desc">Partial specialization for initializing a complex value </td></tr>
+<tr id="row_0_8_2_0_2_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1RandomUniformFunc.html" target="_self">RandomUniformFunc</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_2_0_3_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1RandomUniformFunc_3_01complex_3_01Element_01_4_01_4.html" target="_self">RandomUniformFunc&lt; complex&lt; Element &gt; &gt;</a></td><td class="desc">Partial specialization for initializing a complex value </td></tr>
+<tr id="row_0_8_2_0_4_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorContainsFunc.html" target="_self">TensorContainsFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_5_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorCopyIf.html" target="_self">TensorCopyIf</a></td><td class="desc">Helper to conditionally copy between tensor views </td></tr>
+<tr id="row_0_8_2_0_6_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorEqualsFunc.html" target="_self">TensorEqualsFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_7_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFillDiagonalFunc.html" target="_self">TensorFillDiagonalFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_8_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFillFunc.html" target="_self">TensorFillFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_9_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFillGaussianFunc.html" target="_self">TensorFillGaussianFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_2_0_10_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFillLinearFunc.html" target="_self">TensorFillLinearFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_11_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFillRandomUniformFunc.html" target="_self">TensorFillRandomUniformFunc</a></td><td class="desc">Computes a random Gaussian distribution </td></tr>
+<tr id="row_0_8_2_0_12_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorForEachHelper.html" target="_self">TensorForEachHelper</a></td><td class="desc">Helper to perform for-each operation </td></tr>
+<tr id="row_0_8_2_0_13_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorForEachHelper_3_01Func_00_01Rank_00_010_01_4.html" target="_self">TensorForEachHelper&lt; Func, Rank, 0 &gt;</a></td><td class="desc">Helper to perform for-each operation </td></tr>
+<tr id="row_0_8_2_0_14_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorFuncBinaryOp.html" target="_self">TensorFuncBinaryOp</a></td><td class="desc">Helper to apply a binary operator in place </td></tr>
+<tr id="row_0_8_2_0_15_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TensorUpdateOffDiagonalFunc.html" target="_self">TensorUpdateOffDiagonalFunc</a></td><td class="desc">&lt; Layout function </td></tr>
+<tr id="row_0_8_2_0_16_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1detail_1_1TrivialConvert.html" target="_self">TrivialConvert</a></td><td class="desc">Helper to convert between types </td></tr>
+<tr id="row_0_8_2_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1BlockForEach.html" target="_self">BlockForEach</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_2_2_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1Gemm.html" target="_self">Gemm</a></td><td class="desc"></td></tr>
+<tr id="row_0_8_2_3_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01LayoutB_193dd3a37f00deff1e5dcd7c310afb1f.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, ComputeType, arch::OpMultiplyAdd &gt;</a></td><td class="desc">Partial specialization for multiply-add </td></tr>
+<tr id="row_0_8_2_4_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01LayoutB_55729eac7dbd6bf311ea36f680e83e93.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, ComputeType, arch::OpMultiplyAddSaturate &gt;</a></td><td class="desc">Partial specialization for multiply-add-saturate </td></tr>
+<tr id="row_0_8_2_5_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1reference_1_1host_1_1Gemm_3_01ElementA_00_01LayoutA_00_01ElementB_00_01LayoutB_4f3f32c4b336238abfd741e87bfced46.html" target="_self">Gemm&lt; ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, ScalarType, ComputeType, arch::OpXorPopc &gt;</a></td><td class="desc">Partial specialization for XOR-popc </td></tr>
+<tr id="row_0_9_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_9_" class="arrow" onclick="toggleFolder('0_9_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_9_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1thread_1_1Matrix.html" target="_self">Matrix</a></td><td class="desc">Per-thread matrix object storing a packed matrix </td></tr>
+<tr id="row_0_10_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_10_" class="arrow" onclick="toggleFolder('0_10_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_0_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_0_" class="arrow" onclick="toggleFolder('0_10_0_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1transform_1_1thread.html" target="_self">thread</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_0_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1thread_1_1Transpose.html" target="_self">Transpose</a></td><td class="desc">Transforms a fragment by doing a transpose </td></tr>
+<tr id="row_0_10_0_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1thread_1_1Transpose_3_01ElementCount___00_01layout_1_1PitchLinearS99f8e05faf0bb5ed48a0154afe740d81.html" target="_self">Transpose&lt; ElementCount_, layout::PitchLinearShape&lt; 4, 4 &gt;, int8_t &gt;</a></td><td class="desc">Specialization for int8_t 4x4 transpose </td></tr>
+<tr id="row_0_10_1_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_1_" class="arrow" onclick="toggleFolder('0_10_1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacecutlass_1_1transform_1_1threadblock.html" target="_self">threadblock</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator.html" target="_self">PredicatedTileAccessIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_1_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile.html" target="_self">PredicatedTileAccessIterator2dThreadTile</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_2_" class="arrow" onclick="toggleFolder('0_10_1_2_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__da632779aba661c0f4cfaaa78126b771.html" target="_self">PredicatedTileAccessIterator2dThreadTile&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_2_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__18e9cf25bb3b8edfaad595241a6dc2d7.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_3_" class="arrow" onclick="toggleFolder('0_10_1_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__1790abaa54a01f277d75766d5882fec8.html" target="_self">PredicatedTileAccessIterator2dThreadTile&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_3_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__8ccc62d47a092afc8bee32ffe9d1e4ba.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_4_" class="arrow" onclick="toggleFolder('0_10_1_4_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__7327fa15996bcb8502cdfcc192350fe1.html" target="_self">PredicatedTileAccessIterator2dThreadTile&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_4_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator2dThreadTile_3_01Shape__a56cbccec33ee916292ad9d068474609.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_5_" class="arrow" onclick="toggleFolder('0_10_1_5_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen89c687c583745a73cb485041911a4c4e.html" target="_self">PredicatedTileAccessIterator&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_5_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemenc07b5ec72f83e782121ac629288d61fe.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_6_" class="arrow" onclick="toggleFolder('0_10_1_6_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemenab63a1e105bf37f6371516cb9e2c5a7a.html" target="_self">PredicatedTileAccessIterator&lt; Shape_, Element_, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_6_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemena9b06926a275b569ee9f7f142604b997.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_7_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_7_" class="arrow" onclick="toggleFolder('0_10_1_7_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen784a0e9da3f55064c47e5613791f51f7.html" target="_self">PredicatedTileAccessIterator&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_7_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen41e459f664d17473570cf22fb616845f.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_8_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_8_" class="arrow" onclick="toggleFolder('0_10_1_8_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen9838736ad62fae54213fbaf722a989ab.html" target="_self">PredicatedTileAccessIterator&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_8_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen44ce348364e78f5a56fa0c2cef6af930.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_9_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_9_" class="arrow" onclick="toggleFolder('0_10_1_9_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen809793e785fb4211888c6b4e5dcfcb39.html" target="_self">PredicatedTileAccessIterator&lt; Shape_, Element_, layout::RowMajorInterleaved&lt; InterleavedK &gt;, AdvanceRank, ThreadMap_, AccessType_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_9_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileAccessIterator_3_01Shape___00_01Elemen058417e2cdd86f3cd6ad5458581571c8.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_10_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator.html" target="_self">PredicatedTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_11_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile.html" target="_self">PredicatedTileIterator2dThreadTile</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_12_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_12_" class="arrow" onclick="toggleFolder('0_10_1_12_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_0165b39a630d10785a3558406f9adb99b9.html" target="_self">PredicatedTileIterator2dThreadTile&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, Transpose_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_12_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_01e11ed7192af5d7ad1bce5641fa13112e.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_13_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_13_" class="arrow" onclick="toggleFolder('0_10_1_13_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_017a517f3c73efd795ab05059cc9b111e1.html" target="_self">PredicatedTileIterator2dThreadTile&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, Transpose_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_13_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_0b878062cc0cd214bf7e17d74ff17e246.html" target="_self">AccessType</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_13_1_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_0145ef045e8f7d57dc718098adcb00cf3d.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_14_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_14_" class="arrow" onclick="toggleFolder('0_10_1_14_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_013671177d6219bfeb0e1b4dc4c1b5bf11.html" target="_self">PredicatedTileIterator2dThreadTile&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, Transpose_ &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_14_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator2dThreadTile_3_01Shape___00_0102e766863c6ac9ec2063a02c4803eecb.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_15_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_15_" class="arrow" onclick="toggleFolder('0_10_1_15_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___0068b3e874b5d93d11f0fa902c7f1d11d9.html" target="_self">PredicatedTileIterator&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, AccessSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_15_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___00a6b756b1bcfbb35fe4a3e68ff074e380.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_16_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_16_" class="arrow" onclick="toggleFolder('0_10_1_16_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___00f6b3a9dfab5e7c72d5233f7e5e6e3b9b.html" target="_self">PredicatedTileIterator&lt; Shape_, Element_, layout::ColumnMajorInterleaved&lt; InterleavedK &gt;, AdvanceRank, ThreadMap_, AccessSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_16_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___00ebd1a63351e1085d0b718582ec7b06c8.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_17_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_17_" class="arrow" onclick="toggleFolder('0_10_1_17_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___00e7c2c404e7aedfe60ad56bb5571306a1.html" target="_self">PredicatedTileIterator&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, AccessSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_17_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___006a5f2f7a8271031e6cdc5daa5441f2af.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_18_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_18_" class="arrow" onclick="toggleFolder('0_10_1_18_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___0041ea81994f8af0d4d071fdb9e66b5ff0.html" target="_self">PredicatedTileIterator&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, AccessSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_18_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___004d0f9b5e19c29acc17bcdc360dafebbd.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_19_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_19_" class="arrow" onclick="toggleFolder('0_10_1_19_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___00d670f969180a8d182dffb356ebcc957e.html" target="_self">PredicatedTileIterator&lt; Shape_, Element_, layout::RowMajorInterleaved&lt; InterleavedK &gt;, AdvanceRank, ThreadMap_, AccessSize &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_19_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1PredicatedTileIterator_3_01Shape___00_01Element___009fd89f6dad84238fd7d63df0a0c0364f.html" target="_self">Params</a></td><td class="desc">Parameters object is precomputed state and is host-constructible </td></tr>
+<tr id="row_0_10_1_20_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator.html" target="_self">RegularTileAccessIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_21_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__eb7d20f8b9d69e0ae5e7ef51dc480867.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_22_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__2c1476eaf582bfe972793e17babfe985.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::ColumnMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_23_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__a3c11cf1f00ef7a1efb8389ac6e4c6e0.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::ColumnMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_24_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__0855e9d9ab619202d2397180c1e4c4a5.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_25_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__f04332958a49a47d6fb2b25201764630.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_26_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__6baada077236f1a368c61c5e11b45b72.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::RowMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_27_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__0184b7188941788a96624510a4b2f876.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::RowMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_28_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_28_" class="arrow" onclick="toggleFolder('0_10_1_28_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__ebf4714349612673e8b6609b763eeb6f.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::TensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_28_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element_0a9491607d11be8e1780e79ad711aa42.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_29_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_29_" class="arrow" onclick="toggleFolder('0_10_1_29_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element__e9a9e0f4286f652f55eb9b863b21effe.html" target="_self">RegularTileAccessIterator&lt; Shape_, Element_, layout::TensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_29_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileAccessIterator_3_01Shape___00_01Element_3be8b96d170d886f39b6b30acab65e7a.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_30_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator.html" target="_self">RegularTileIterator</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_31_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator2dThreadTile.html" target="_self">RegularTileIterator2dThreadTile</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_32_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator2dThreadTile_3_01Shape___00_01Eleb60d066756d1c18f05fceee6a27bdb8a.html" target="_self">RegularTileIterator2dThreadTile&lt; Shape_, Element_, layout::ColumnMajorInterleaved&lt; 4 &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for interleaved layout + 2d thread-tiled threadmapping </td></tr>
+<tr id="row_0_10_1_33_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator2dThreadTile_3_01Shape___00_01Ele76ed82829532ae1c17f4c78158f036c7.html" target="_self">RegularTileIterator2dThreadTile&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for pitch-linear + 2d thread-tiled threadmapping </td></tr>
+<tr id="row_0_10_1_34_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator2dThreadTile_3_01Shape___00_01Ele654c8f6161ae5340f040397a4e2e045c.html" target="_self">RegularTileIterator2dThreadTile&lt; Shape_, Element_, layout::RowMajorInterleaved&lt; 4 &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for interleaved layout + 2d thread-tiled threadmapping </td></tr>
+<tr id="row_0_10_1_35_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_011d3637dbd8bc58bcb020b51bf57fbfc0.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajor, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for pitch-linear </td></tr>
+<tr id="row_0_10_1_36_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_017982f81d4ef592e19c8427de2ea933a3.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_37_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_010889a732373c350de9b9a9f6c13cd761.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_38_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01187f8574e1fe9d7d5e8fbf09bd834bf0.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajorVoltaTensorOpMultiplicandBCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_39_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01793f74bfd8f116a827948ab01a37349a.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajorVoltaTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_40_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01bd31b3810c1fedf2e7e5959ff92b5d3d.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::ColumnMajorVoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Shape_::kRow &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_41_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_0184a89653916f5d51ab59d1b386989a17.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::PitchLinear, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for pitch-linear </td></tr>
+<tr id="row_0_10_1_42_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_0149454d361ea5885cf5166a920b5145df.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajor, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc">Regular tile iterator specialized for pitch-linear </td></tr>
+<tr id="row_0_10_1_43_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01c20d35180520077a5a09b1e33543c1a5.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajorTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_44_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01a31b454d9c930525c1e9ca406a514f40.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajorTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_45_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_0104ad31bd559a88cc418ae1cab7492ed5.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajorVoltaTensorOpMultiplicandBCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_46_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01f6f6511b5033cad31083644ac69c54d8.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajorVoltaTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_47_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01b3fa5720e807697de61b9f937b269cd0.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::RowMajorVoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Shape_::kColumn &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_48_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_48_" class="arrow" onclick="toggleFolder('0_10_1_48_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01efd5013a2503d6567e2bf6b40c97360c.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::TensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value, int(128/sizeof(Element_))&gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_48_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_052caec9d5bceeb59b9a13cb3338ce64d.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_49_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_49_" class="arrow" onclick="toggleFolder('0_10_1_49_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_0197fef2242a3454a7d1cebe61aee28b43.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::TensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Crosswise &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_49_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_039093927f4b1ee61538c569bf1ae4efd.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_50_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_50_" class="arrow" onclick="toggleFolder('0_10_1_50_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01a75d2cd74e722d6ad6a3b41aabfd432d.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::VoltaTensorOpMultiplicandBCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_50_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_02d305cfb0b55c6fb236a52cf2240651e.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_51_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_51_" class="arrow" onclick="toggleFolder('0_10_1_51_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01f96bbeb63e6d4ce4a2551279de3a9f0e.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::VoltaTensorOpMultiplicandCongruous&lt; sizeof_bits&lt; Element_ &gt;::value &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_51_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_032f88d1be8b209e44a4815c707ba35bb.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_1_52_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span id="arr_0_10_1_52_" class="arrow" onclick="toggleFolder('0_10_1_52_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_01dbd6b8468d5bd787308d2f615a24d123.html" target="_self">RegularTileIterator&lt; Shape_, Element_, layout::VoltaTensorOpMultiplicandCrosswise&lt; sizeof_bits&lt; Element_ &gt;::value, Shape_::kContiguous &gt;, AdvanceRank, ThreadMap_, Alignment &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_1_52_0_" style="display:none;"><td class="entry"><span style="width:80px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1threadblock_1_1RegularTileIterator_3_01Shape___00_01Element___00_0390833403016f5d817416e20828845df.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection </td></tr>
+<tr id="row_0_10_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinear2DThreadTileStripminedThreadMap.html" target="_self">PitchLinear2DThreadTileStripminedThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_3_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_3_" class="arrow" onclick="toggleFolder('0_10_3_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinear2DThreadTileStripminedThreadMap_3_01Shape___00_01Thread0082c3467229b12cc9dd996283ee7160.html" target="_self">PitchLinear2DThreadTileStripminedThreadMap&lt; Shape_, Threads, cutlass::layout::PitchLinearShape&lt; 4, 4 &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_3_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinear2DThreadTileStripminedThreadMap_3_01Shape___00_01Thread896c01a3c466da1bf392e0cdfced4d53.html" target="_self">Detail</a></td><td class="desc">Internal implementation details </td></tr>
+<tr id="row_0_10_4_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_4_" class="arrow" onclick="toggleFolder('0_10_4_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearStripminedThreadMap.html" target="_self">PitchLinearStripminedThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_4_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearStripminedThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc">Internal implementation details </td></tr>
+<tr id="row_0_10_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearTilePolicyStripminedThreadContiguous.html" target="_self">PitchLinearTilePolicyStripminedThreadContiguous</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_6_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearTilePolicyStripminedThreadStrided.html" target="_self">PitchLinearTilePolicyStripminedThreadStrided</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_7_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_7_" class="arrow" onclick="toggleFolder('0_10_7_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearWarpRakedThreadMap.html" target="_self">PitchLinearWarpRakedThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_7_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearWarpRakedThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection Iterations along each dimension (concept: PitchLinearShape) </td></tr>
+<tr id="row_0_10_8_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_8_" class="arrow" onclick="toggleFolder('0_10_8_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearWarpStripedThreadMap.html" target="_self">PitchLinearWarpStripedThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_8_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1PitchLinearWarpStripedThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection Iterations along each dimension (concept: PitchLinearShape) </td></tr>
+<tr id="row_0_10_9_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span id="arr_0_10_9_" class="arrow" onclick="toggleFolder('0_10_9_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1TransposePitchLinearThreadMap.html" target="_self">TransposePitchLinearThreadMap</a></td><td class="desc"></td></tr>
+<tr id="row_0_10_9_0_" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1TransposePitchLinearThreadMap_1_1Detail.html" target="_self">Detail</a></td><td class="desc">Internal details made public to facilitate introspection Iterations along each dimension (concept: PitchLinearShape) </td></tr>
+<tr id="row_0_10_10_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1TransposePitchLinearThreadMap2DThreadTile.html" target="_self">TransposePitchLinearThreadMap2DThreadTile</a></td><td class="desc">Thread Mapping a 2D threadtiled mapping as a transposed Pitchlinear2DThreadTile mapping </td></tr>
+<tr id="row_0_10_11_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1transform_1_1TransposePitchLinearThreadMapSimt.html" target="_self">TransposePitchLinearThreadMapSimt</a></td><td class="desc"></td></tr>
+<tr id="row_0_11_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1AlignedArray.html" target="_self">AlignedArray</a></td><td class="desc">Aligned array type </td></tr>
+<tr id="row_0_12_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1AlignedBuffer.html" target="_self">AlignedBuffer</a></td><td class="desc">Modifies semantics of cutlass::Array&lt;&gt; to provide guaranteed alignment </td></tr>
+<tr id="row_0_13_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_13_" class="arrow" onclick="toggleFolder('0_13_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4.html" target="_self">Array&lt; T, N, false &gt;</a></td><td class="desc">Statically sized array for any data type </td></tr>
+<tr id="row_0_13_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1const__iterator.html" target="_self">const_iterator</a></td><td class="desc">Bidirectional constant iterator over elements </td></tr>
+<tr id="row_0_13_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1const__reference.html" target="_self">const_reference</a></td><td class="desc">Reference object extracts sub-byte items </td></tr>
+<tr id="row_0_13_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1const__reverse__iterator.html" target="_self">const_reverse_iterator</a></td><td class="desc">Bidirectional constant iterator over elements </td></tr>
+<tr id="row_0_13_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1iterator.html" target="_self">iterator</a></td><td class="desc">Bidirectional iterator over elements </td></tr>
+<tr id="row_0_13_4_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1reference.html" target="_self">reference</a></td><td class="desc">Reference object inserts or extracts sub-byte items </td></tr>
+<tr id="row_0_13_5_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01false_01_4_1_1reverse__iterator.html" target="_self">reverse_iterator</a></td><td class="desc">Bidirectional iterator over elements </td></tr>
+<tr id="row_0_14_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_14_" class="arrow" onclick="toggleFolder('0_14_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4.html" target="_self">Array&lt; T, N, true &gt;</a></td><td class="desc">Statically sized array for any data type </td></tr>
+<tr id="row_0_14_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__iterator.html" target="_self">const_iterator</a></td><td class="desc">Bidirectional constant iterator over elements </td></tr>
+<tr id="row_0_14_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__reverse__iterator.html" target="_self">const_reverse_iterator</a></td><td class="desc">Bidirectional constant iterator over elements </td></tr>
+<tr id="row_0_14_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1iterator.html" target="_self">iterator</a></td><td class="desc">Bidirectional iterator over elements </td></tr>
+<tr id="row_0_14_3_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1reverse__iterator.html" target="_self">reverse_iterator</a></td><td class="desc">Bidirectional iterator over elements </td></tr>
+<tr id="row_0_15_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1CommandLine.html" target="_self">CommandLine</a></td><td class="desc"></td></tr>
+<tr id="row_0_16_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1complex.html" target="_self">complex</a></td><td class="desc"></td></tr>
+<tr id="row_0_17_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1ConstSubbyteReference.html" target="_self">ConstSubbyteReference</a></td><td class="desc"></td></tr>
+<tr id="row_0_18_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1Coord.html" target="_self">Coord</a></td><td class="desc">Statically-sized array specifying Coords within a tensor </td></tr>
+<tr id="row_0_19_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1cuda__exception.html" target="_self">cuda_exception</a></td><td class="desc">C++ exception wrapper for CUDA <code>cudaError_t</code> </td></tr>
+<tr id="row_0_20_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1Distribution.html" target="_self">Distribution</a></td><td class="desc"><a class="el" href="structcutlass_1_1Distribution.html" title="Distribution type. ">Distribution</a> type </td></tr>
+<tr id="row_0_21_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1divide__assert.html" target="_self">divide_assert</a></td><td class="desc"></td></tr>
+<tr id="row_0_22_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1divides.html" target="_self">divides</a></td><td class="desc"></td></tr>
+<tr id="row_0_23_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1divides_3_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">divides&lt; Array&lt; half_t, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_24_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1divides_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">divides&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_25_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1FloatType.html" target="_self">FloatType</a></td><td class="desc">Defines a floating-point type based on the number of exponent and mantissa bits </td></tr>
+<tr id="row_0_26_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1FloatType_3_0111_00_0152_01_4.html" target="_self">FloatType&lt; 11, 52 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_27_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1FloatType_3_015_00_0110_01_4.html" target="_self">FloatType&lt; 5, 10 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_28_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1FloatType_3_018_00_0123_01_4.html" target="_self">FloatType&lt; 8, 23 &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_29_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1half__t.html" target="_self">half_t</a></td><td class="desc">IEEE half-precision floating-point type </td></tr>
+<tr id="row_0_30_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1HostTensor.html" target="_self">HostTensor</a></td><td class="desc">Host tensor </td></tr>
+<tr id="row_0_31_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1IdentityTensorLayout.html" target="_self">IdentityTensorLayout</a></td><td class="desc"></td></tr>
+<tr id="row_0_32_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1integer__subbyte.html" target="_self">integer_subbyte</a></td><td class="desc">4-bit signed integer type </td></tr>
+<tr id="row_0_33_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType.html" target="_self">IntegerType</a></td><td class="desc">Defines integers based on size and whether they are signed </td></tr>
+<tr id="row_0_34_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_011_00_01false_01_4.html" target="_self">IntegerType&lt; 1, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_35_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_011_00_01true_01_4.html" target="_self">IntegerType&lt; 1, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_36_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0116_00_01false_01_4.html" target="_self">IntegerType&lt; 16, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_37_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0116_00_01true_01_4.html" target="_self">IntegerType&lt; 16, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_38_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0132_00_01false_01_4.html" target="_self">IntegerType&lt; 32, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_39_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0132_00_01true_01_4.html" target="_self">IntegerType&lt; 32, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_40_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_014_00_01false_01_4.html" target="_self">IntegerType&lt; 4, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_41_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_014_00_01true_01_4.html" target="_self">IntegerType&lt; 4, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_42_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0164_00_01false_01_4.html" target="_self">IntegerType&lt; 64, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_43_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_0164_00_01true_01_4.html" target="_self">IntegerType&lt; 64, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_44_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_018_00_01false_01_4.html" target="_self">IntegerType&lt; 8, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_45_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1IntegerType_3_018_00_01true_01_4.html" target="_self">IntegerType&lt; 8, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_46_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1is__pow2.html" target="_self">is_pow2</a></td><td class="desc"></td></tr>
+<tr id="row_0_47_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1KernelLaunchConfiguration.html" target="_self">KernelLaunchConfiguration</a></td><td class="desc">Structure containing the basic launch configuration of a CUDA kernel </td></tr>
+<tr id="row_0_48_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1log2__down.html" target="_self">log2_down</a></td><td class="desc"></td></tr>
+<tr id="row_0_49_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1log2__down_3_01N_00_011_00_01Count_01_4.html" target="_self">log2_down&lt; N, 1, Count &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_50_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1log2__up.html" target="_self">log2_up</a></td><td class="desc"></td></tr>
+<tr id="row_0_51_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1log2__up_3_01N_00_011_00_01Count_01_4.html" target="_self">log2_up&lt; N, 1, Count &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_52_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1MatrixCoord.html" target="_self">MatrixCoord</a></td><td class="desc"></td></tr>
+<tr id="row_0_53_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1MatrixShape.html" target="_self">MatrixShape</a></td><td class="desc">Describes the size of a matrix tile </td></tr>
+<tr id="row_0_54_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1Max.html" target="_self">Max</a></td><td class="desc"></td></tr>
+<tr id="row_0_55_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1maximum.html" target="_self">maximum</a></td><td class="desc"></td></tr>
+<tr id="row_0_56_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1maximum_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">maximum&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_57_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1maximum_3_01float_01_4.html" target="_self">maximum&lt; float &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_58_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1Min.html" target="_self">Min</a></td><td class="desc"></td></tr>
+<tr id="row_0_59_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minimum.html" target="_self">minimum</a></td><td class="desc"></td></tr>
+<tr id="row_0_60_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minimum_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">minimum&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_61_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minimum_3_01float_01_4.html" target="_self">minimum&lt; float &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_62_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minus.html" target="_self">minus</a></td><td class="desc"></td></tr>
+<tr id="row_0_63_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minus_3_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">minus&lt; Array&lt; half_t, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_64_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1minus_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">minus&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_65_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiplies.html" target="_self">multiplies</a></td><td class="desc"></td></tr>
+<tr id="row_0_66_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiplies_3_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">multiplies&lt; Array&lt; half_t, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_67_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiplies_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">multiplies&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_68_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add.html" target="_self">multiply_add</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_69_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add_3_01Array_3_01half__t_00_01N_01_4_00_01Array_3_01half__t_00_01N_01adaeadb27c0e4439444709c0eb30963.html" target="_self">multiply_add&lt; Array&lt; half_t, N &gt;, Array&lt; half_t, N &gt;, Array&lt; half_t, N &gt; &gt;</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_70_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add_3_01Array_3_01T_00_01N_01_4_00_01Array_3_01T_00_01N_01_4_00_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">multiply_add&lt; Array&lt; T, N &gt;, Array&lt; T, N &gt;, Array&lt; T, N &gt; &gt;</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_71_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add_3_01complex_3_01T_01_4_00_01complex_3_01T_01_4_00_01complex_3_01T_01_4_01_4.html" target="_self">multiply_add&lt; complex&lt; T &gt;, complex&lt; T &gt;, complex&lt; T &gt; &gt;</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_72_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add_3_01complex_3_01T_01_4_00_01T_00_01complex_3_01T_01_4_01_4.html" target="_self">multiply_add&lt; complex&lt; T &gt;, T, complex&lt; T &gt; &gt;</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_73_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1multiply__add_3_01T_00_01complex_3_01T_01_4_00_01complex_3_01T_01_4_01_4.html" target="_self">multiply_add&lt; T, complex&lt; T &gt;, complex&lt; T &gt; &gt;</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_0_74_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1negate.html" target="_self">negate</a></td><td class="desc"></td></tr>
+<tr id="row_0_75_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1negate_3_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">negate&lt; Array&lt; half_t, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_76_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1negate_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">negate&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_77_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericArrayConverter.html" target="_self">NumericArrayConverter</a></td><td class="desc">Conversion operator for Array </td></tr>
+<tr id="row_0_78_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericArrayConverter_3_01float_00_01half__t_00_012_00_01Round_01_4.html" target="_self">NumericArrayConverter&lt; float, half_t, 2, Round &gt;</a></td><td class="desc">Partial specialization for Array&lt;float, 2&gt; &lt;= Array&lt;half_t, 2&gt;, round to nearest </td></tr>
+<tr id="row_0_79_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericArrayConverter_3_01float_00_01half__t_00_01N_00_01Round_01_4.html" target="_self">NumericArrayConverter&lt; float, half_t, N, Round &gt;</a></td><td class="desc">Partial specialization for Array&lt;half&gt; &lt;= Array&lt;float&gt; </td></tr>
+<tr id="row_0_80_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericArrayConverter_3_01half__t_00_01float_00_012_00_01FloatRoundStyle_1_1round__to__nearest_01_4.html" target="_self">NumericArrayConverter&lt; half_t, float, 2, FloatRoundStyle::round_to_nearest &gt;</a></td><td class="desc">Partial specialization for Array&lt;half, 2&gt; &lt;= Array&lt;float, 2&gt;, round to nearest </td></tr>
+<tr id="row_0_81_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericArrayConverter_3_01half__t_00_01float_00_01N_00_01Round_01_4.html" target="_self">NumericArrayConverter&lt; half_t, float, N, Round &gt;</a></td><td class="desc">Partial specialization for Array&lt;half&gt; &lt;= Array&lt;float&gt; </td></tr>
+<tr id="row_0_82_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter.html" target="_self">NumericConverter</a></td><td class="desc"></td></tr>
+<tr id="row_0_83_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter_3_01float_00_01half__t_00_01Round_01_4.html" target="_self">NumericConverter&lt; float, half_t, Round &gt;</a></td><td class="desc">Partial specialization for float &lt;= <a class="el" href="structcutlass_1_1half__t.html" title="IEEE half-precision floating-point type. ">half_t</a> </td></tr>
+<tr id="row_0_84_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter_3_01half__t_00_01float_00_01FloatRoundStyle_1_1round__to__nearest_01_4.html" target="_self">NumericConverter&lt; half_t, float, FloatRoundStyle::round_to_nearest &gt;</a></td><td class="desc">Specialization for round-to-nearest </td></tr>
+<tr id="row_0_85_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter_3_01half__t_00_01float_00_01FloatRoundStyle_1_1round__toward__zero_01_4.html" target="_self">NumericConverter&lt; half_t, float, FloatRoundStyle::round_toward_zero &gt;</a></td><td class="desc">Specialization for round-toward-zero </td></tr>
+<tr id="row_0_86_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter_3_01int8__t_00_01float_00_01Round_01_4.html" target="_self">NumericConverter&lt; int8_t, float, Round &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_87_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverter_3_01T_00_01T_00_01Round_01_4.html" target="_self">NumericConverter&lt; T, T, Round &gt;</a></td><td class="desc">Partial specialization for float &lt;= <a class="el" href="structcutlass_1_1half__t.html" title="IEEE half-precision floating-point type. ">half_t</a> </td></tr>
+<tr id="row_0_88_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1NumericConverterClamp.html" target="_self">NumericConverterClamp</a></td><td class="desc"></td></tr>
+<tr id="row_0_89_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1plus.html" target="_self">plus</a></td><td class="desc"></td></tr>
+<tr id="row_0_90_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1plus_3_01Array_3_01half__t_00_01N_01_4_01_4.html" target="_self">plus&lt; Array&lt; half_t, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_91_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1plus_3_01Array_3_01T_00_01N_01_4_01_4.html" target="_self">plus&lt; Array&lt; T, N &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_92_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_92_" class="arrow" onclick="toggleFolder('0_92_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1PredicateVector.html" target="_self">PredicateVector</a></td><td class="desc">Statically sized array of bits implementing </td></tr>
+<tr id="row_0_92_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1PredicateVector_1_1ConstIterator.html" target="_self">ConstIterator</a></td><td class="desc">An iterator implementing <a class="el" href="group__predicate__iterator__concept.html">Predicate Iterator Concept</a> enabling sequential read and write access to predicates </td></tr>
+<tr id="row_0_92_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1PredicateVector_1_1Iterator.html" target="_self">Iterator</a></td><td class="desc">An iterator implementing <a class="el" href="group__predicate__iterator__concept.html">Predicate Iterator Concept</a> enabling sequential read and write access to predicates </td></tr>
+<tr id="row_0_92_2_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1PredicateVector_1_1TrivialIterator.html" target="_self">TrivialIterator</a></td><td class="desc"><a class="el" href="classcutlass_1_1PredicateVector_1_1Iterator.html" title="An iterator implementing Predicate Iterator Concept enabling sequential read and write access to pred...">Iterator</a> that always returns true </td></tr>
+<tr id="row_0_93_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1RealType.html" target="_self">RealType</a></td><td class="desc">Used to determine the real-valued underlying type of a numeric type T </td></tr>
+<tr id="row_0_94_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1RealType_3_01complex_3_01T_01_4_01_4.html" target="_self">RealType&lt; complex&lt; T &gt; &gt;</a></td><td class="desc">Partial specialization for complex-valued type </td></tr>
+<tr id="row_0_95_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1ReferenceFactory.html" target="_self">ReferenceFactory</a></td><td class="desc"></td></tr>
+<tr id="row_0_96_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1ReferenceFactory_3_01Element_00_01false_01_4.html" target="_self">ReferenceFactory&lt; Element, false &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_97_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1ReferenceFactory_3_01Element_00_01true_01_4.html" target="_self">ReferenceFactory&lt; Element, true &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_98_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1ScalarIO.html" target="_self">ScalarIO</a></td><td class="desc">Helper to enable formatted printing of CUTLASS scalar types to an ostream </td></tr>
+<tr id="row_0_99_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1Semaphore.html" target="_self">Semaphore</a></td><td class="desc">CTA-wide semaphore for inter-CTA synchronization </td></tr>
+<tr id="row_0_100_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits.html" target="_self">sizeof_bits</a></td><td class="desc">Defines the size of an element in bits </td></tr>
+<tr id="row_0_101_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits_3_01Array_3_01T_00_01N_00_01RegisterSized_01_4_01_4.html" target="_self">sizeof_bits&lt; Array&lt; T, N, RegisterSized &gt; &gt;</a></td><td class="desc">Statically sized array for any data type </td></tr>
+<tr id="row_0_102_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits_3_01bin1__t_01_4.html" target="_self">sizeof_bits&lt; bin1_t &gt;</a></td><td class="desc">Defines the size of an element in bits - specialized for bin1_t </td></tr>
+<tr id="row_0_103_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits_3_01int4b__t_01_4.html" target="_self">sizeof_bits&lt; int4b_t &gt;</a></td><td class="desc">Defines the size of an element in bits - specialized for int4b_t </td></tr>
+<tr id="row_0_104_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits_3_01uint1b__t_01_4.html" target="_self">sizeof_bits&lt; uint1b_t &gt;</a></td><td class="desc">Defines the size of an element in bits - specialized for uint1b_t </td></tr>
+<tr id="row_0_105_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sizeof__bits_3_01uint4b__t_01_4.html" target="_self">sizeof_bits&lt; uint4b_t &gt;</a></td><td class="desc">Defines the size of an element in bits - specialized for uint4b_t </td></tr>
+<tr id="row_0_106_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1sqrt__est.html" target="_self">sqrt_est</a></td><td class="desc"></td></tr>
+<tr id="row_0_107_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1SubbyteReference.html" target="_self">SubbyteReference</a></td><td class="desc"></td></tr>
+<tr id="row_0_108_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1Tensor4DCoord.html" target="_self">Tensor4DCoord</a></td><td class="desc">Defines a canonical 4D coordinate used by tensor operations </td></tr>
+<tr id="row_0_109_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1TensorRef.html" target="_self">TensorRef</a></td><td class="desc"></td></tr>
+<tr id="row_0_110_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classcutlass_1_1TensorView.html" target="_self">TensorView</a></td><td class="desc"></td></tr>
+<tr id="row_0_111_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits.html" target="_self">TypeTraits</a></td><td class="desc"></td></tr>
+<tr id="row_0_112_" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span id="arr_0_112_" class="arrow" onclick="toggleFolder('0_112_')">&#9658;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01double_01_4_01_4.html" target="_self">TypeTraits&lt; complex&lt; double &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_112_0_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01double_01_4_01_4_1_1integer__type.html" target="_self">integer_type</a></td><td class="desc"></td></tr>
+<tr id="row_0_112_1_" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01double_01_4_01_4_1_1unsigned__type.html" target="_self">unsigned_type</a></td><td class="desc"></td></tr>
+<tr id="row_0_113_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01float_01_4_01_4.html" target="_self">TypeTraits&lt; complex&lt; float &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_114_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01half_01_4_01_4.html" target="_self">TypeTraits&lt; complex&lt; half &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_115_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01complex_3_01half__t_01_4_01_4.html" target="_self">TypeTraits&lt; complex&lt; half_t &gt; &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_116_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01double_01_4.html" target="_self">TypeTraits&lt; double &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_117_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01float_01_4.html" target="_self">TypeTraits&lt; float &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_118_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01half__t_01_4.html" target="_self">TypeTraits&lt; half_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_119_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01int_01_4.html" target="_self">TypeTraits&lt; int &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_120_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01int64__t_01_4.html" target="_self">TypeTraits&lt; int64_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_121_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01int8__t_01_4.html" target="_self">TypeTraits&lt; int8_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_122_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01uint64__t_01_4.html" target="_self">TypeTraits&lt; uint64_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_123_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01uint8__t_01_4.html" target="_self">TypeTraits&lt; uint8_t &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_124_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1TypeTraits_3_01unsigned_01_4.html" target="_self">TypeTraits&lt; unsigned &gt;</a></td><td class="desc"></td></tr>
+<tr id="row_0_125_" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structcutlass_1_1xor__add.html" target="_self">xor_add</a></td><td class="desc">Fused multiply-add </td></tr>
+<tr id="row_1_"><td class="entry"><span style="width:0px;display:inline-block;">&#160;</span><span id="arr_1_" class="arrow" onclick="toggleFolder('1_')">&#9658;</span><span class="icona"><span class="icon">N</span></span><b>std</b></td><td class="desc">STL namespace </td></tr>
+<tr id="row_1_0_" class="even" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structstd_1_1numeric__limits_3_01cutlass_1_1half__t_01_4.html" target="_self">numeric_limits&lt; cutlass::half_t &gt;</a></td><td class="desc">Numeric limits </td></tr>
+<tr id="row_2_" class="even"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structDebugType.html" target="_self">DebugType</a></td><td class="desc"></td></tr>
+<tr id="row_3_"><td class="entry"><span style="width:16px;display:inline-block;">&#160;</span><span class="icona"><span class="icon">C</span></span><a class="el" href="structDebugValue.html" target="_self">DebugValue</a></td><td class="desc"></td></tr>
+</table>
+</div><!-- directory -->
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma_8h.html

@@ -0,0 +1,156 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: mma.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li><li class="navelem"><a class="el" href="dir_048c1df36ab9c2efbb0733edba6291c9.html">arch</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">arch/mma.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Templates exposing architecture support for multiply-add operations.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &quot;<a class="el" href="array_8h_source.html">cutlass/array.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="numeric__types_8h_source.html">cutlass/numeric_types.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="include_2cutlass_2gemm_2gemm_8h_source.html">cutlass/gemm/gemm.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="arch_2mma__sm50_8h_source.html">cutlass/arch/mma_sm50.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="arch_2mma__sm60_8h_source.html">cutlass/arch/mma_sm60.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="arch_2mma__sm61_8h_source.html">cutlass/arch/mma_sm61.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="mma__sm70_8h_source.html">cutlass/arch/mma_sm70.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="mma__sm75_8h_source.html">cutlass/arch/mma_sm75.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for arch/mma.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma_8h__incl.png" border="0" usemap="#mma_8h" alt=""/></div>
+<map name="mma_8h" id="mma_8h">
+</map>
+</div>
+</div><div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma_8h__dep__incl.png" border="0" usemap="#mma_8hdep" alt=""/></div>
+<map name="mma_8hdep" id="mma_8hdep">
+</map>
+</div>
+</div>
+<p><a href="arch_2mma_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma.html">cutlass::arch::Mma&lt; Shape_, kThreads_, ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, Operator &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01ElementAb6e65b2cf5ede7f41cb070a767158dee.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, ElementA, LayoutA, ElementB, LayoutB, ElementC, LayoutC, Operator &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation - specialized for 1x1x1x1 matrix multiply operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01ElementAb6e65b2cf5ede7f41cb070a767158dee.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:namespacecutlass_1_1arch"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass_1_1arch.html">cutlass::arch</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma_8h__dep__incl.md5

@@ -0,0 +1 @@
+7d16b59e6ba0442b8a275a213d5da3a6

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma_8h__incl.md5

@@ -0,0 +1 @@
+d1fff3f9d55a262110aa6a456caa91e0

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm50_8h.html

@@ -0,0 +1,176 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: mma_sm50.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li><li class="navelem"><a class="el" href="dir_048c1df36ab9c2efbb0733edba6291c9.html">arch</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">arch/mma_sm50.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Matrix multiply.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &quot;<a class="el" href="arch_2mma_8h_source.html">cutlass/arch/mma.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="complex_8h_source.html">cutlass/complex.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="layout_2matrix_8h_source.html">cutlass/layout/matrix.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="include_2cutlass_2gemm_2gemm_8h_source.html">cutlass/gemm/gemm.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for arch/mma_sm50.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm50_8h__incl.png" border="0" usemap="#mma__sm50_8h" alt=""/></div>
+<map name="mma__sm50_8h" id="mma__sm50_8h">
+</map>
+</div>
+</div><div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm50_8h__dep__incl.png" border="0" usemap="#mma__sm50_8hdep" alt=""/></div>
+<map name="mma__sm50_8hdep" id="mma__sm50_8hdep">
+</map>
+</div>
+</div>
+<p><a href="arch_2mma__sm50_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_004bb3fd76ca2af7b3210676fa9644d95b.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, float, LayoutA, float, LayoutB, float, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_004bb3fd76ca2af7b3210676fa9644d95b.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_0aa57e6a2e6b5da37d10688bf99419a23.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, double, LayoutA, double, LayoutB, double, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_0aa57e6a2e6b5da37d10688bf99419a23.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01int_00_00b2dff9ce8caad9aff5bc6a355539161.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, int, LayoutA, int, LayoutB, int, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01int_00_00b2dff9ce8caad9aff5bc6a355539161.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_76f9d24016e1b4167b16f4d7628c9546.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; float &gt;, LayoutA, complex&lt; float &gt;, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_76f9d24016e1b4167b16f4d7628c9546.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_f1c9d2ee842455cd0c5b71d56108d468.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; float &gt;, LayoutA, float, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_f1c9d2ee842455cd0c5b71d56108d468.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_00e3e12e263df6506b8cf06c3f4d478b8e.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, float, LayoutA, complex&lt; float &gt;, LayoutB, complex&lt; float &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01float_00e3e12e263df6506b8cf06c3f4d478b8e.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_30fa42e1ad201df010637cd22fc070a1.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; double &gt;, LayoutA, complex&lt; double &gt;, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_30fa42e1ad201df010637cd22fc070a1.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_48b3a43bc03fff93a111ac01abe7e40d.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, complex&lt; double &gt;, LayoutA, double, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01complex_48b3a43bc03fff93a111ac01abe7e40d.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_070b94670e040ed5855e5b42d5ca8a443.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, double, LayoutA, complex&lt; double &gt;, LayoutB, complex&lt; double &gt;, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01double_070b94670e040ed5855e5b42d5ca8a443.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01half__t_4f30ee91f7bb3844ff7579c68d078818.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, float, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_011_01_4_00_011_00_01half__t_4f30ee91f7bb3844ff7579c68d078818.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:namespacecutlass_1_1arch"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass_1_1arch.html">cutlass::arch</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm50_8h__dep__incl.md5

@@ -0,0 +1 @@
+988e6466c703c4e63c9a889b8c3c54b5

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm50_8h__incl.md5

@@ -0,0 +1 @@
+03f1613fdffbd6e7575de0d2967d08bf

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm60_8h.html

@@ -0,0 +1,157 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: mma_sm60.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li><li class="navelem"><a class="el" href="dir_048c1df36ab9c2efbb0733edba6291c9.html">arch</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">arch/mma_sm60.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Matrix multiply.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &lt;cuda_fp16.h&gt;</code><br />
+<code>#include &quot;<a class="el" href="arch_2mma_8h_source.html">cutlass/arch/mma.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="layout_2matrix_8h_source.html">cutlass/layout/matrix.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for arch/mma_sm60.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm60_8h__incl.png" border="0" usemap="#mma__sm60_8h" alt=""/></div>
+<map name="mma__sm60_8h" id="mma__sm60_8h">
+</map>
+</div>
+</div><div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm60_8h__dep__incl.png" border="0" usemap="#mma__sm60_8hdep" alt=""/></div>
+<map name="mma__sm60_8hdep" id="mma__sm60_8hdep">
+</map>
+</div>
+</div>
+<p><a href="arch_2mma__sm60_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_011_00_011_01_4_00_011_00_01half__t_8cf78649807b93684f3d431bfa34ee28.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 2, 1, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, half_t, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_011_00_011_01_4_00_011_00_01half__t_8cf78649807b93684f3d431bfa34ee28.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_012_00_011_01_4_00_011_00_01half__t_f3dc2e59f857ada163d1e0781ea8f391.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 2, 1 &gt;, 1, half_t, LayoutA, half_t, LayoutB, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_012_00_011_01_4_00_011_00_01half__t_f3dc2e59f857ada163d1e0781ea8f391.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_ccde11d1bbbdab3702772ce44eb9729a.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 2, 2, 1 &gt;, 1, half_t, layout::ColumnMajor, half_t, layout::RowMajor, half_t, layout::ColumnMajor, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_ccde11d1bbbdab3702772ce44eb9729a.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_c07cc6439298fa5486a719e577be2538.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 2, 2, 1 &gt;, 1, half_t, layout::ColumnMajor, half_t, layout::RowMajor, half_t, layout::RowMajor, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_012_00_012_00_011_01_4_00_011_00_01half__t_c07cc6439298fa5486a719e577be2538.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:namespacecutlass_1_1arch"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass_1_1arch.html">cutlass::arch</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm60_8h__dep__incl.md5

@@ -0,0 +1 @@
+ba69b14e3936946092854211499ae9fa

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm60_8h__incl.md5

@@ -0,0 +1 @@
+e820099c55f2397639bb210d76ec4c05

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm61_8h.html

@@ -0,0 +1,149 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: mma_sm61.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li><li class="navelem"><a class="el" href="dir_048c1df36ab9c2efbb0733edba6291c9.html">arch</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">arch/mma_sm61.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Matrix multiply.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &quot;<a class="el" href="layout_2matrix_8h_source.html">cutlass/layout/matrix.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for arch/mma_sm61.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm61_8h__incl.png" border="0" usemap="#mma__sm61_8h" alt=""/></div>
+<map name="mma__sm61_8h" id="mma__sm61_8h">
+</map>
+</div>
+</div><div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_2mma__sm61_8h__dep__incl.png" border="0" usemap="#mma__sm61_8hdep" alt=""/></div>
+<map name="mma__sm61_8hdep" id="mma__sm61_8hdep">
+</map>
+</div>
+</div>
+<p><a href="arch_2mma__sm61_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_014_01_4_00_011_00_01int8__t_a1ef6624fc8c10126f17f4ee88283d72.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 4 &gt;, 1, int8_t, LayoutA, int8_t, LayoutB, int, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_014_01_4_00_011_00_01int8__t_a1ef6624fc8c10126f17f4ee88283d72.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_012_01_4_00_011_00_01int16__t8c4bac365710598317a69c489f7239db.html">cutlass::arch::Mma&lt; gemm::GemmShape&lt; 1, 1, 2 &gt;, 1, int16_t, layout::RowMajor, int16_t, layout::ColumnMajor, int, LayoutC, OpMultiplyAdd &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Matrix multiply-add operation.  <a href="structcutlass_1_1arch_1_1Mma_3_01gemm_1_1GemmShape_3_011_00_011_00_012_01_4_00_011_00_01int16__t8c4bac365710598317a69c489f7239db.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:namespacecutlass_1_1arch"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass_1_1arch.html">cutlass::arch</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm61_8h__dep__incl.md5

@@ -0,0 +1 @@
+1faaf1631d5f0e44d6cc6c7121e6972e

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_2mma__sm61_8h__incl.md5

@@ -0,0 +1 @@
+8cce8aef2d98c4082d68734b538253c7

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_8h.html

@@ -0,0 +1,147 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: arch.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li><li class="navelem"><a class="el" href="dir_048c1df36ab9c2efbb0733edba6291c9.html">arch</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a>  </div>
+  <div class="headertitle">
+<div class="title">arch.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Defines tags for architecture-specific configurations.  
+<a href="#details">More...</a></p>
+<div class="textblock"><div class="dynheader">
+This graph shows which files directly or indirectly include this file:</div>
+<div class="dyncontent">
+<div class="center"><img src="arch_8h__dep__incl.png" border="0" usemap="#arch_8hdep" alt=""/></div>
+<map name="arch_8hdep" id="arch_8hdep">
+</map>
+</div>
+</div>
+<p><a href="arch_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm50.html">cutlass::arch::Sm50</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm60.html">cutlass::arch::Sm60</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm61.html">cutlass::arch::Sm61</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm70.html">cutlass::arch::Sm70</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm72.html">cutlass::arch::Sm72</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1arch_1_1Sm75.html">cutlass::arch::Sm75</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:namespacecutlass_1_1arch"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass_1_1arch.html">cutlass::arch</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/arch_8h__dep__incl.md5

@@ -0,0 +1 @@
+9ea32ea41ab87776449ab855965480b3

crates/vendor/candle/candle-flash-attn/cutlass/docs/array_8h.html

@@ -0,0 +1,167 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.11"/>
+<title>CUTLASS: array.h File Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<script type="text/javascript">
+  $(document).ready(function() { init_search(); });
+</script>
+<script type="text/x-mathjax-config">
+  MathJax.Hub.Config({
+    extensions: ["tex2jax.js"],
+    jax: ["input/TeX","output/HTML-CSS"],
+});
+</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+  <td id="projectlogo"><img alt="Logo" src="cutlass-logo-small.png"/></td>
+  <td id="projectalign" style="padding-left: 0.5em;">
+   <div id="projectname">CUTLASS
+   </div>
+   <div id="projectbrief">CUDA Templates for Linear Algebra Subroutines and Solvers</div>
+  </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.11 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+  <div id="navrow1" class="tabs">
+    <ul class="tablist">
+      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
+      <li><a href="namespaces.html"><span>Namespaces</span></a></li>
+      <li><a href="annotated.html"><span>Classes</span></a></li>
+      <li class="current"><a href="files.html"><span>Files</span></a></li>
+      <li>
+        <div id="MSearchBox" class="MSearchBoxInactive">
+        <span class="left">
+          <img id="MSearchSelect" src="search/mag_sel.png"
+               onmouseover="return searchBox.OnSearchSelectShow()"
+               onmouseout="return searchBox.OnSearchSelectHide()"
+               alt=""/>
+          <input type="text" id="MSearchField" value="Search" accesskey="S"
+               onfocus="searchBox.OnSearchFieldFocus(true)" 
+               onblur="searchBox.OnSearchFieldFocus(false)" 
+               onkeyup="searchBox.OnSearchFieldChange(event)"/>
+          </span><span class="right">
+            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
+          </span>
+        </div>
+      </li>
+    </ul>
+  </div>
+  <div id="navrow2" class="tabs2">
+    <ul class="tablist">
+      <li><a href="files.html"><span>File&#160;List</span></a></li>
+      <li><a href="globals.html"><span>File&#160;Members</span></a></li>
+    </ul>
+  </div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+     onmouseover="return searchBox.OnSearchSelectShow()"
+     onmouseout="return searchBox.OnSearchSelectHide()"
+     onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0" 
+        name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+  <ul>
+<li class="navelem"><a class="el" href="dir_d44c64559bbebec7f509842c48db8b23.html">include</a></li><li class="navelem"><a class="el" href="dir_6baf2bb612a2f0daa69af3101ede80a1.html">cutlass</a></li>  </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+  <div class="summary">
+<a href="#nested-classes">Classes</a> &#124;
+<a href="#namespaces">Namespaces</a> &#124;
+<a href="#func-members">Functions</a>  </div>
+  <div class="headertitle">
+<div class="title">array.h File Reference</div>  </div>
+</div><!--header-->
+<div class="contents">
+
+<p>Statically sized array of elements that accommodates all CUTLASS-supported numeric types and is safe to use in a union.  
+<a href="#details">More...</a></p>
+<div class="textblock"><code>#include &quot;<a class="el" href="cutlass_8h_source.html">cutlass/cutlass.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="numeric__types_8h_source.html">cutlass/numeric_types.h</a>&quot;</code><br />
+<code>#include &quot;<a class="el" href="array__subbyte_8h_source.html">cutlass/array_subbyte.h</a>&quot;</code><br />
+</div><div class="textblock"><div class="dynheader">
+Include dependency graph for array.h:</div>
+<div class="dyncontent">
+<div class="center"><img src="array_8h__incl.png" border="0" usemap="#array_8h" alt=""/></div>
+<map name="array_8h" id="array_8h">
+</map>
+</div>
+</div>
+<p><a href="array_8h_source.html">Go to the source code of this file.</a></p>
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="nested-classes"></a>
+Classes</h2></td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">struct &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="structcutlass_1_1sizeof__bits_3_01Array_3_01T_00_01N_00_01RegisterSized_01_4_01_4.html">cutlass::sizeof_bits&lt; Array&lt; T, N, RegisterSized &gt; &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Statically sized array for any data type.  <a href="structcutlass_1_1sizeof__bits_3_01Array_3_01T_00_01N_00_01RegisterSized_01_4_01_4.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4.html">cutlass::Array&lt; T, N, true &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Statically sized array for any data type.  <a href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1iterator.html">cutlass::Array&lt; T, N, true &gt;::iterator</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Bidirectional iterator over elements.  <a href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1iterator.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__iterator.html">cutlass::Array&lt; T, N, true &gt;::const_iterator</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Bidirectional constant iterator over elements.  <a href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__iterator.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1reverse__iterator.html">cutlass::Array&lt; T, N, true &gt;::reverse_iterator</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Bidirectional iterator over elements.  <a href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1reverse__iterator.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__reverse__iterator.html">cutlass::Array&lt; T, N, true &gt;::const_reverse_iterator</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Bidirectional constant iterator over elements.  <a href="classcutlass_1_1Array_3_01T_00_01N_00_01true_01_4_1_1const__reverse__iterator.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:"><td class="memItemLeft" align="right" valign="top">class &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classcutlass_1_1AlignedArray.html">cutlass::AlignedArray&lt; T, N, Alignment &gt;</a></td></tr>
+<tr class="memdesc:"><td class="mdescLeft">&#160;</td><td class="mdescRight">Aligned array type.  <a href="classcutlass_1_1AlignedArray.html#details">More...</a><br /></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="namespaces"></a>
+Namespaces</h2></td></tr>
+<tr class="memitem:namespacecutlass"><td class="memItemLeft" align="right" valign="top"> &#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html">cutlass</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table><table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a>
+Functions</h2></td></tr>
+<tr class="memitem:a935aabfdc47cf03f87c67bb22533f97f"><td class="memItemLeft" align="right" valign="top"><a class="el" href="cutlass_8h.html#a28c2443a142676d3d71effdae1a986b1">CUTLASS_HOST_DEVICE</a> <a class="el" href="platform_8h.html#a72f0657181cca64b44eb186b707eb380">constexpr</a> bool&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html#a935aabfdc47cf03f87c67bb22533f97f">cutlass::ispow2</a> (unsigned x)</td></tr>
+<tr class="memdesc:a935aabfdc47cf03f87c67bb22533f97f"><td class="mdescLeft">&#160;</td><td class="mdescRight">Returns true if the argument is a power of 2.  <a href="namespacecutlass.html#a935aabfdc47cf03f87c67bb22533f97f">More...</a><br /></td></tr>
+<tr class="separator:a935aabfdc47cf03f87c67bb22533f97f"><td class="memSeparator" colspan="2">&#160;</td></tr>
+<tr class="memitem:ac16d8caf23537912eb02123c4bdacd14"><td class="memItemLeft" align="right" valign="top"><a class="el" href="cutlass_8h.html#a28c2443a142676d3d71effdae1a986b1">CUTLASS_HOST_DEVICE</a> <a class="el" href="platform_8h.html#a72f0657181cca64b44eb186b707eb380">constexpr</a> unsigned&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="namespacecutlass.html#ac16d8caf23537912eb02123c4bdacd14">cutlass::floor_pow_2</a> (unsigned x)</td></tr>
+<tr class="memdesc:ac16d8caf23537912eb02123c4bdacd14"><td class="mdescLeft">&#160;</td><td class="mdescRight">Returns the largest power of two not greater than the argument.  <a href="namespacecutlass.html#ac16d8caf23537912eb02123c4bdacd14">More...</a><br /></td></tr>
+<tr class="separator:ac16d8caf23537912eb02123c4bdacd14"><td class="memSeparator" colspan="2">&#160;</td></tr>
+</table>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by &#160;<a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.11
+</small></address>
+</body>
+</html>

crates/vendor/candle/candle-flash-attn/cutlass/docs/array_8h__incl.md5

@@ -0,0 +1 @@
+90c159bd7ad938ad2d6e263ea8402fe7