CANN/asc-devkit矩阵计算优化实践
Matrix Compute Practices Sample Introduction
【免费下载链接】asc-devkit本项目是CANN 推出的昇腾AI处理器专用的算子程序开发语言，原生支持C和C++标准规范，主要由类库和语言扩展层构成，提供多层级API，满足多维场景算子开发诉求。项目地址: https://gitcode.com/cann/asc-devkit
Overview
Matrix computation optimization samples based on Matrix Compute API, introducing Matmul and MxFP4 Matmul high-performance practices in high-level API and basic API scenarios through<<<>>>direct call mode.
Sample List
| Directory Name | Function Description |
|---|---|
| matmul_high_performance | Matmul high-level API progressive performance optimization sample, demonstrating multi-core splitting, MDL, L1/L2 Cache, constant tiling, UnitFlag, and other optimization methods. |
| matmul_basic_api_high_performance | Matmul basic API best practices sample, based on static Tensor programming demonstrating basic API high-performance implementation details. |
| matmul_mxfp4_high_performance | MxFP4 Matmul high-level API performance tuning sample, demonstrating constant tiling and scale data transfer optimization methods. |
| matmul_mxfp4_basic_api_high_performance | MxFP4 Matmul basic API high-performance sample, based on static Tensor programming demonstrating verified basic API implementation paths. |
【免费下载链接】asc-devkit本项目是CANN 推出的昇腾AI处理器专用的算子程序开发语言，原生支持C和C++标准规范，主要由类库和语言扩展层构成，提供多层级API，满足多维场景算子开发诉求。项目地址: https://gitcode.com/cann/asc-devkit
创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考