github/ROCm
1
1
Fork 0
mirror of https://github.com/ROCm/ROCm.git synced 2026-04-05 03:01:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

Skip BLOCK_SIZE that is too large compare to M/N (#449)

Browse Source
This commit is contained in:
Lixun Zhang
2024-01-09 13:41:09 -06:00
committed by GitHub
parent c2766bbd5f
commit ce9dacec72
1 changed files with 3 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
scripts/amd/gemm/tune_gemm.py
View File

@@ -68,6 +68,9 @@ def prune_configs(M, N, K, configs):
continue
if N <= mfma and BLOCK_SIZE_N != mfma:
continue
# Skip BLOCK_SIZE that is too large compare to M/N
if BLOCK_SIZE_M > M * 2 or BLOCK_SIZE_N > N * 2:
continue
# skip large split_k when not necessary
if SPLIT_K != 1 and not need_split_k(M, N, K):
continue