[codex] Optimize hidden-size 512 RMSNorm dispatch (#24710)

Co-authored-by: Codex <codex@example.com>
This commit is contained in:
Xiaoyu Zhang
2026-05-19 09:26:10 +08:00
committed by GitHub
co-authored by Codex
parent 170fe57cf0
commit 2424303dfb
3 changed files with 33 additions and 21 deletions
@@ -88,7 +88,7 @@ def test_rmsnorm_hidden_size_support(hidden_size: int) -> None:
(64, "RMSNormWarpKernel"),
(128, "RMSNormWarpKernel"),
(256, "RMSNormWarpKernel"),
(512, "RMSNormKernel"),
(512, "RMSNormHalfKernel"),
(1536, "RMSNormKernel"),
(2048, "RMSNormHalfKernel"),
(2304, "RMSNormKernel"), # NOTE: not 512 aligned