OwnerSunshine530 0128fb8100 perf: Triton kernel 两个dot改fp16 Tensor Core(flash标准:fp16 matmul+fp32 acc),单块提速2-4x ...

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-17 00:36:25 +08:00

..

code

perf: Triton kernel 两个dot改fp16 Tensor Core(flash标准:fp16 matmul+fp32 acc),单块提速2-4x

2026-06-17 00:36:25 +08:00

main.ipynb

chore: 更新 notebook 环境和文档

2026-06-12 20:51:14 +08:00

任务提交接口说明.md

chore: 初始化 CTI 推理优化项目

2026-06-03 13:49:30 +08:00