[Triton-MLIR] Increase block size K to completely eliminate shared memory bank conflicts (#862)

2022-11-08 17:39:23 -08:00
parent 080b4addf8
commit 2da71b2aaa
2 changed files with 14 additions and 8 deletions
--- a/python/tutorials/03-matrix-multiplication.py
+++ b/python/tutorials/03-matrix-multiplication.py
@@ -156,7 +156,7 @@ import triton.language as tl

@triton.autotune(
    configs=[
-        triton.Config({'BLOCK_SIZE_M': 128, 'BLOCK_SIZE_N': 256, 'BLOCK_SIZE_K': 32, 'GROUP_SIZE_M': 8}, num_stages=3, num_warps=8),
+        triton.Config({'BLOCK_SIZE_M': 128, 'BLOCK_SIZE_N': 256, 'BLOCK_SIZE_K': 64, 'GROUP_SIZE_M': 8}, num_stages=3, num_warps=8),
    ],
    key=['M', 'N', 'K'],
 )