heavenluo

heavenluo pushed to my-work at heavenluo/op_optimization

2026-08-02 13:32:15 +08:00

3047133520 1

f8ec53d590 docs(fused_moe): 新增面试版优化问题复盘与负收益实录（learnlog）

e2d3e84774 fused_moe: MACA C++ 优化全过程（895=89.5 基线 + m4 async + 4 个同步变体实测）

Compare 3 commits »

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-26 00:13:57 +08:00

71f7235030 fix: ako4x-run-maca-cpp 断链 gitlink 改为普通跟踪文件

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-25 23:21:49 +08:00

ef4b163e82 sync: 将 learnlog 迁入「基于AI Agent…」目录并跟踪 read/ 工作文档

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-25 22:56:47 +08:00

5dddfdcad6 1

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-24 19:22:22 +08:00

fc9effb42f 添加 read/ 目录下的 txt 参考文档

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-24 19:09:17 +08:00

3ea246e438 添加 read/ 目录下的 md 参考文档

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-24 18:31:15 +08:00

9d15d5b22e 1

d73f557f2f Fix MACA native loader ABI and shape hook

Compare 2 commits »

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-23 23:01:30 +08:00

4c94bf6017 read

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-23 14:36:17 +08:00

fbfba3f771 添加 VSCode RemoteSSH 远程代理与Codex配置指南（learnlog）

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-14 18:10:44 +08:00

977e0593e3 chore: 纳入 ako4x_c500 参赛工作树快照到赛题目录

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-14 16:13:52 +08:00

feb785a9b0 gitignore: 忽略 read/ 大文件参考资料（含 622M zip，勿入仓库）

cd4616f7f2 添加 fused_moe / AKO4X 学习笔记（learnlog）

Compare 2 commits »

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-12 23:55:15 +08:00

bcaa8bb70d 添加 ako4x 子模块

heavenluo created branch my-work in heavenluo/op_optimization

2026-07-12 21:42:50 +08:00

heavenluo pushed to my-work at heavenluo/op_optimization

2026-07-12 21:42:50 +08:00

heavenluo created repository heavenluo/op_optimization

2026-07-12 21:22:57 +08:00

heavenluo created pull request ccf-ai-infra/Intro-ops#28

2026-05-28 00:45:58 +08:00

补充了四个算子的tilelang端和nvidia端的实现代码

heavenluo pushed to submit-intro-ops at heavenluo/Intro-ops

2026-05-28 00:09:02 +08:00

6d2d463c06 address review comments

heavenluo pushed to submit-intro-ops at heavenluo/Intro-ops

2026-05-28 00:05:06 +08:00

heavenluo created branch submit-intro-ops in heavenluo/Intro-ops

2026-05-28 00:05:04 +08:00

heavenluo pushed to dev at heavenluo/Intro-ops

2026-05-27 22:48:12 +08:00