biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Bugfix] Fix Maverick correctness by filling zero to cache space in cutlass_moe (#20167)

Browse Source 
Signed-off-by: Ming Yang <yming@meta.com>
This commit is contained in:
Ming Yang
2025-07-07 18:07:22 -07:00
committed by GitHub
parent d2e841a10a
commit afb7cff1b9
2 changed files with 120 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/model_executor/layers/fused_moe/cutlass_moe.py
View File

@@ -180,7 +180,11 @@ def run_cutlass_moe_fp8(
c2 = _resize_cache(workspace2, (M * topk, N))
c3 = _resize_cache(workspace13, (M * topk, K))
c1.fill_(0)
if not per_act_token and (expert_map is not None or use_batched_format):
# this is necessary to avoid imprecise scale calculation caused by
# random data in the unused workspace. The workspace is unused when
# this rank handles only partial tokens, or when it is batched .
c1.fill_(0)
ops.cutlass_moe_mm(c1, a1q, w1, a1q_scale, w1_scale, expert_offsets,
problem_sizes1, ab_strides1, ab_strides1, c_strides1,
@@ -303,7 +307,7 @@ class CutlassExpertsFp8(mk.FusedMoEPermuteExpertsUnpermute):
):
assert w1_zp is None, "w1_zp is not supported in CUTLASS MoE"
assert w2_zp is None, "w2_zp is not supported in CUTLASS MoE"
activation_callable = lambda i, o: self.activation(activation, i, o)
activation_callable = lambda o, i: self.activation(activation, o, i)
in_dtype = hidden_states.dtype
run_cutlass_moe_fp8(
output, hidden_states, w1, w2, topk_ids, activation_callable,