kernels/moe test pruning (#27053)

Signed-off-by: Fardin Hoque <kfhfar@amazon.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
Fardin Hoque
2025-10-29 21:10:34 -07:00
committed by GitHub
parent 17d055f527
commit b8c48c5d72
13 changed files with 34 additions and 56 deletions

View File

@@ -130,10 +130,8 @@ def run_single_case(m, n, k, topk, num_experts, block_size):
# Note: N <= 512 will disable the deepgemm path due to performance issues.
MNKs = [
(1024, 768, 128),
(1024, 768, 512),
(2048, 768, 512),
(512, 1024, 1024),
(512, 2048, 2048),
(4096, 4096, 1024),
]