example/ck_tile/05_reduce/reduce.cpp · 0b8f117f1ae765f0e490368de7f0c5d7591b17b6 · gaoqiong / composable_kernel_ROCM · GitLab

Find file Blame History Permalink

[CK_TILE] Adjust kBlockSize of reduce example for better perf (#1779) · 0b8f117f
ClementLinCF authored Jan 13, 2025
```
* Observed a 2x perf improvement with kBlockSize = 256
* Using 512 threads may lead to redundant computations
```
0b8f117f

reduce.cpp 3.92 KB