cubecl-convolution 0.8.1

CubeCL Convolution Kernels Engine
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

//! Readers read from global memory and write to stage memory
//!
//! Readers fall into two axes:
//!
//! - **Synchronization**
//!   - **Synchronous**: Performs direct memory accesses.
//!   - **Asynchronous**: Uses `memcpy_async` for loading.
//!
//! - **Coverage**
//!   - **Full**: Loads the entire shared memory region.
//!   - **Partial**: Loads only a single stage, when multiple stages share the same memory.

mod reader;

pub use reader::*;