ZCG12345 3be8e33fe9 [Kernel] Add moe_gating_top_k operator support for Ascend NPU (#5579) ...

### What this PR does / why we need it?

1.replace moe_gating_top_k from torch_npu with custom op
2.enable the  renorm function of moe_gating_top_k in softmax scenerio

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
No need test

- vLLM version: v0.13.0
- vLLM main:
7157596103

---------

Signed-off-by: ZCG12345 <2097562023@qq.com>

2026-01-07 21:42:31 +08:00

..

aclnn_torch_adapter

[feature] Add Custom Op grouped_matmul_swiglu_quant (#4431)

2025-11-27 21:56:18 +08:00

batch_matmul_transpose

[Bugfix]fix bmm_transpose ops for cann version (#4653)

2025-12-06 10:52:46 +08:00

cmake

add dispatch_gmm_combine kernel (#3532)

2025-12-04 23:00:59 +08:00

dispatch_ffn_combine

[bugfix] Fix dummy-run and multi-node issues in MoE routing and MTP (#4947)

2025-12-15 14:18:23 +08:00

dispatch_gmm_combine_decode

[Feature]EPLB:Adapt DispatchGmmCombineDecode operator to eplb tensor list and expert token numbers (#5552)

2026-01-07 11:23:42 +08:00

dispatch_layout

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00

grouped_matmul_swiglu_quant_weight_nz_tensor_list

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

kernels

[BugFix]Fix precision issue for LoRA feature (#4141)

2025-12-19 14:22:06 +08:00

lightning_indexer

[kernel] add AscendC op: lightning_indexer and sparse_flash_attention (#4625)

2025-12-03 09:53:10 +08:00

matmul_allreduce_add_rmsnorm

[BugFix][kernel] fix matmul_allreduce_add_rmsnorm_kernel (#5335)

2026-01-05 15:19:54 +08:00

mla_preprocess

[feat] mlapo add bf16 no_quant support (#4852)

2025-12-11 11:06:56 +08:00

moe_combine_normal

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00

moe_dispatch_normal

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00

moe_gating_top_k

[Kernel] Add moe_gating_top_k operator support for Ascend NPU (#5579)

2026-01-07 21:42:31 +08:00

moe_init_routing_custom

[OP] add custom op aclnnMoeInitRoutingCustom (#5251)

2025-12-29 19:29:40 +08:00

notify_dispatch

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00

sparse_flash_attention

[kernel] add AscendC op: lightning_indexer and sparse_flash_attention (#4625)

2025-12-03 09:53:10 +08:00

third_party

add dispatch_gmm_combine kernel (#3532)

2025-12-04 23:00:59 +08:00

utils

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00

build_aclnn.sh

[Kernel] Add moe_gating_top_k operator support for Ascend NPU (#5579)

2026-01-07 21:42:31 +08:00

build.sh

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

camem_allocator.cpp

Quality enhancement: Immediately interrupt execution when memory OOM (#3932)

2025-11-04 08:55:09 +08:00

CMakeLists.txt

[Kernel]update csrc cmakelist for open-source cann (#5458)

2025-12-29 20:34:53 +08:00

ops.h

mlapo add qdown output (#4707)

2025-12-06 11:18:53 +08:00

torch_binding_meta.cpp

[Kernel] Add moe_gating_top_k operator support for Ascend NPU (#5579)

2026-01-07 21:42:31 +08:00

torch_binding.cpp

[Kernel] Add moe_gating_top_k operator support for Ascend NPU (#5579)

2026-01-07 21:42:31 +08:00

utils.h

[Kernel] Add moe normal ops (#4810)

2025-12-10 17:15:28 +08:00