Composable Kernel User Guide#
This document contains instructions for installing, using, and contributing to Composable Kernel (CK).
Composable Kernel (CK) library aims to provide a programming model for writing performance critical kernels for machine learning workloads across multiple architectures including GPUs, CPUs, etc, through general purpose kernel languages, like HIP C++.
CK utilizes two concepts to achieve performance portability and code maintainability:
A tile-based programming model
Algorithm complexity reduction for complex ML operators, using innovative technique we call “Tensor Coordinate Transformation”.
Current CK library are structured into 4 layers:
“Templated Tile Operators” layer
“Templated Kernel and Invoker” layer
“Instantiated Kernel and Invoker” layer
“Client API” layer
The following is a list of CK documents in the suggested reading order:
- 1. CK Hello world
- 2. CK docker hub
- 3. Supported Primitives Guide
- 4. API Reference Guide
- 5. Contributor’s Guide