vllm/vllm/model_executor/layers/fused_moe at db01535e2bd5641e8cd7f88e5febf1d2a26bfbce - vllm

configs

[Perf] Add tuned triton moe config for Qwen3.5 H200, 9.9% E2E throughput improvement (#37340 )

2026-03-18 14:18:34 -04:00

experts

Revert "[MoE Kernel] Flashinfer nvfp4 cutedsl moe kernel integration" (#38050 ) (#38169 )

2026-03-26 07:59:09 -07:00

oracle

Revert "[MoE Kernel] Flashinfer nvfp4 cutedsl moe kernel integration" (#38050 ) (#38169 )

2026-03-26 07:59:09 -07:00

prepare_finalize

[MoE] Move PF Methods to Folder (#35927 )

2026-03-22 02:42:59 -06:00

router

[Bugfix] Register VLLM_BATCH_INVARIANT in envs.py to fix spurious unknown env var warning (#35007 )

2026-03-23 18:31:14 -04:00

runner

[MoE Refactor] DefaultMoERunner simplifcation (#33049 )

2026-03-19 15:07:44 -04:00

__init__.py

[MoE Refactor] Create MK for TRTLLM Kernels (#32564 )

2026-03-03 10:39:50 -08:00

activation.py

[Refactor] Replace activation: str with MoEActivation enum (#33843 )

2026-02-11 17:29:32 -08:00

all2all_utils.py

[MoE] Move PF Methods to Folder (#35927 )

2026-03-22 02:42:59 -06:00

batched_deep_gemm_moe.py

[ROCm] Fix MoE kernel test failures on gfx950 (#37833 )

2026-03-25 13:46:40 -05:00

config.py

[Bugfix][ROCm][MoE] Fix mxfp4 oracle regressions from #37128 (#37787 )

2026-03-25 08:17:33 +08:00

cpu_fused_moe.py

[Torch 2.11] Guard torch._C._cpu attribute checks for forward compatibility (#35673 )

2026-03-17 18:47:59 +00:00

cutlass_moe.py

[Kernel] Add non-gated support for NVFP4 CUTLASS MoE (#37320 )

2026-03-17 18:12:04 -04:00

deep_gemm_moe.py

[Kernel] Add FlashInfer MoE A2A Kernel (#36022 )

2026-03-15 23:45:32 -07:00

deep_gemm_utils.py

[Bugfix] ep_scatter kernel store-load race condition (#34991 )

2026-03-13 01:07:59 +00:00

fallback.py

[Feature]: Remove Chunking From FusedMoE (#34086 )

2026-03-12 14:24:38 -04:00

flashinfer_cutlass_moe.py

Fix eplb nvfp4 experts hook (#37217 )

2026-03-16 22:03:54 +00:00

flashinfer_trtllm_moe.py

[Bugfix] Disable FlashInfer TRTLLM BF16 path for non-gated MoE (#36146 )

2026-03-05 09:49:05 -08:00

fused_batched_moe.py

[ROCm] Enable DeepEP ROCm as all2allbackend for AMD GPUs. (#34692 )

2026-03-21 00:32:31 -07:00

fused_marlin_moe.py

[MoE Refactor] Mxfp4 oracle rebased (#37128 )

2026-03-21 03:37:04 +00:00

fused_moe_method_base.py

[Perf] Eliminate padding and slicing op for GPT-OSS with Flashinfer MXFP4 MXFP8 MoE (#30647 )

2026-03-18 15:01:26 +00:00

fused_moe_modular_method.py

[MoE Refactor] Create MK for TRTLLM Kernels (#32564 )

2026-03-03 10:39:50 -08:00

fused_moe.py

[Bugfix] Register VLLM_BATCH_INVARIANT in envs.py to fix spurious unknown env var warning (#35007 )

2026-03-23 18:31:14 -04:00

gpt_oss_triton_kernels_moe.py

[ROCm] Fix MoE kernel test failures on gfx950 (#37833 )

2026-03-25 13:46:40 -05:00

layer.py

[Bugfix][ROCm][MoE] Fix mxfp4 oracle regressions from #37128 (#37787 )

2026-03-25 08:17:33 +08:00

modular_kernel.py

[Bugfix][ROCm][MoE] Fix mxfp4 oracle regressions from #37128 (#37787 )

2026-03-25 08:17:33 +08:00

moe_align_block_size.py

[Kernel][MoE] optimize moe_align_block_size (#29642 )

2025-12-07 01:58:47 -08:00

moe_permute_unpermute.py

[Refactor] Remove align block size logic in moe_permute (#33449 )

2026-02-06 10:57:06 -08:00

mori_prepare_finalize.py

[Bugfix][ROCm] Fix MoRI + AITER FP8 dispatch compatibility for defer_input_quant (#37418 )

2026-03-19 09:49:27 +00:00

nixl_ep_prepare_finalize.py

[2/N] Elastic EP Milestone 2: Integrating NIXL-EP (#35627 )

2026-03-13 09:25:33 -04:00

rocm_aiter_fused_moe.py

[Bugfix][ROCm][MoE] Fix mxfp4 oracle regressions from #37128 (#37787 )

2026-03-25 08:17:33 +08:00

routed_experts_capturer.py

[Bugfix] fix device_name for routing replay (#34336 )

2026-02-26 12:18:38 +00:00

shared_fused_moe.py

[MoE Refactor] Introduce MoERunner abstraction and move execution logic from FusedMoE to DefaultMoERunner (#32344 )

2026-02-10 19:51:07 -05:00

topk_weight_and_reduce.py

[MoE Refactor] Create MK for TRTLLM Kernels (#32564 )

2026-03-03 10:39:50 -08:00

triton_cutlass_moe.py

[MoE Refactor] Create MK for TRTLLM Kernels (#32564 )

2026-03-03 10:39:50 -08:00

triton_deep_gemm_moe.py

[MoE Refactor] Create MK for TRTLLM Kernels (#32564 )

2026-03-03 10:39:50 -08:00

unquantized_fused_moe_method.py

[Hardware] Replace torch.cuda.empty_cache with torch.accelerator.empty_cache (#30681 )

2026-03-04 09:49:47 +00:00

utils.py

[ROCm] Enable DeepEP ROCm as all2allbackend for AMD GPUs. (#34692 )

2026-03-21 00:32:31 -07:00

xpu_fused_moe.py

[XPU][MoE Refactor] Refactor xpu mxfp4 support into oracle (#37784 )

2026-03-23 11:10:41 +00:00

zero_expert_fused_moe.py

[MoE Refactor][5/N] Isolate zero expert to LongCatFlash (#28891 )

2025-12-20 18:22:04 +00:00