[Perf] zeros -> empty to remove additional fill (#42988)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-05-21 16:00:29 -04:00
committed by GitHub
parent 9b54e50e2c
commit b29cbf0652
4 changed files with 4 additions and 5 deletions
@@ -255,7 +255,7 @@ def silu_mul_quant_fp8_packed_triton(
if output_q is None:
output_q = torch.empty((M, N_2), dtype=fp8_dtype, device=input.device)
output_scale_packed = torch.zeros(
output_scale_packed = torch.empty(
(num_packed_groups, tma_aligned_M),
dtype=torch.int32,
device=input.device,