metax-maca
/
op_optimization
0
0
Fork 34
Code Issues Pull Requests 4 Packages Projects Releases Wiki Activity

添加开源仓库 #48

Merged
Beckylu merged 1 commits from xiao_xiao/op_optimization:master into master 2026-06-24 15:25:24 +08:00
Conversation 0 Commits 1 Files Changed 1 +5
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
基于AI Agent开发范式的国产GPU大模型推理算子库优化/FlashAttention关键算子迁移与优化.md
View File

@ -255,6 +255,11 @@
* **Tensor Core:** 现代 GPU(含沐曦 C500)的矩阵乘法专用单元,要求维度对齐为 8 或 16 的倍数。
### 5.5 开源仓库参考
[GitHub - MetaX-MACA/flashattn · GitHub](https://github.com/MetaX-MACA/flashattn)
> 链接内容可供用于学习 API、算子实现思路、benchmark 方法和优化策略。选手仍需根据 XPU-OJ 题包接口**自行实现**可提交的 `run_kernel(...)`
---