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

[Bugfix] Disable monolithic TRTLLM MoE for Renormalize routing (#37591) (#37605)

Browse Source 
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>
This commit is contained in:
Vadim Gimpelson
2026-03-20 23:19:26 +04:00
committed by GitHub
parent 12fd17eb51
commit 4f16ebbbd3
5 changed files with 42 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
vllm/model_executor/layers/fused_moe/experts/trtllm_fp8_moe.py
View File

@@ -269,9 +269,16 @@ class TrtLlmFp8ExpertsMonolithic(TrtLlmFp8ExpertsBase, mk.FusedMoEExpertsMonolit
weight_key: QuantKey | None,
activation_key: QuantKey | None,
) -> bool:
"""Monolithic kernels need to express router support."""
"""Monolithic kernels need to express router support.
Renormalize/RenormalizeNaive are excluded: the monolithic kernel's
internal routing for these methods produces output uncorrelated
with the modular kernel's output and with Triton kernel's output
for Qwen3.5-35B-A3B-FP8.
See: https://github.com/vllm-project/vllm/issues/37591
"""
# NOTE(dbari): TopK routing could also be enabled, but need to validate models
# NOTE(dbari): Default is not implemented and should not be enabled until it is
if (weight_key, activation_key) in [
(kFp8Static128BlockSym, kFp8Dynamic128Sym),
(kMxfp8Static, kMxfp8Dynamic),
@@ -279,16 +286,12 @@ class TrtLlmFp8ExpertsMonolithic(TrtLlmFp8ExpertsBase, mk.FusedMoEExpertsMonolit
# NOTE(rob): potentially allow others here. This is a conservative list.
return routing_method in [
RoutingMethodType.DeepSeekV3,
RoutingMethodType.Renormalize,
RoutingMethodType.RenormalizeNaive,
]
elif (weight_key, activation_key) == (kFp8StaticTensorSym, kFp8StaticTensorSym):
# NOTE(dbari): as above, potentially allow others here.
return routing_method in [
RoutingMethodType.DeepSeekV3,
RoutingMethodType.Llama4,
RoutingMethodType.Renormalize,
RoutingMethodType.RenormalizeNaive,
]
else:
raise ValueError("Unsupported quantization scheme.")