Skip to content

GitLab

Unverified Commit b581b225 authored by Cheng Wan's avatar Cheng Wan Committed by GitHub
Browse files

Fix one bug in the grouped-gemm triton kernel (#6772)

parent 69dd878b
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
python/sglang/srt/layers/moe/ep_moe/kernels.py
View file @ b581b225
...@@ -621,7 +621,7 @@ def grouped_gemm_triton_kernel( ...@@ -621,7 +621,7 @@ def grouped_gemm_triton_kernel(
b_ptr += BLOCK_SIZE_K b_ptr += BLOCK_SIZE_K
if use_fp8_w8a8 and not (group_k > 0 and group_n > 0): if use_fp8_w8a8 and not (group_k > 0 and group_n > 0):
scale_a_value = tl.load(scale_a + expert_id) scale_a_value = tl.load(scale_a + m_range_start + offs_am[:, None])
scale_b_value = tl.load(scale_b + expert_id) scale_b_value = tl.load(scale_b + expert_id)
accumulator *= scale_a_value * scale_b_value accumulator *= scale_a_value * scale_b_value
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment