Triton Ascend Case Elemwise Broadcast 2d

2D广播除法优化:小维度完整处理不切分(循环外加载复用),通过固定NUM_BLOCKS实现核间并行(40核),核内SUB_M控制粒度平衡UB利用率,适用于broadcast轴大但非broadcast轴小的2D场景

wenyi-li Updated 10 repo stars

File contents

wenyi-li/awesome-agent-kernel-skills commit dff20ff78e

Frequently asked questions

npx get-skill add wenyi-li/triton-ascend-case-elemwise-broadcast-2d