4 people 19069bcbd5 FP32 router GEMV optimization (#48335) ...

Signed-off-by: peiyuanz <peiyuanz@inferact.ai>
Signed-off-by: Jee Jee Li <jeejeelee@inferact.ai>
Co-authored-by: peiyuanz <peiyuanz@inferact.ai>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: zhouzhou <zhouzhou@zhouzhoudeMacBook-Pro.local>

2026-07-11 13:07:48 +00:00

..

configs

[Kernel][MoE] Tune block-FP8 fused MoE for low-batch decode (#46642)

2026-06-25 12:38:28 -07:00

experts

[XPU]remove is_xxx from moe class and bump up kernels (#48079)

2026-07-11 09:27:13 +00:00

oracle

[Perf] Integrate TRTLLM BF16 MoE Modular Kernel (#45182)

2026-07-09 01:36:14 +04:00

prepare_finalize

[ROCm][DeepEP] Stabilize high-throughput DBO for DP+EP (#46990)

2026-06-29 14:28:02 -07:00

router

FP32 router GEMV optimization (#48335)

2026-07-11 13:07:48 +00:00

runner

[Perf] Minimax M3 - Support cross-layer allreduce-norm fusion (#47631)

2026-07-07 21:16:32 -07:00

__init__.py

[MoE Refactor] FusedMoE/MoERunner inversion refactor (#41184)

2026-06-08 10:42:58 -04:00

activation.py

[Model] Add MiniMax M3 support (#45381)

2026-06-16 01:01:25 +08:00

all2all_utils.py

[Feature] Detect all2all peer fault with fault tolerance backend and prevent corrupted output (#43637)

2026-06-30 09:00:25 -07:00

config.py

[Feature] Support sequence parallel without the need for DP, 1.9%~5.0% E2E Throughput Improvement (#47070)

2026-07-05 05:41:30 -07:00

cpu_fused_moe.py

[Bugfix] Use native SiLU activation in CPU fused MoE (#45961)

2026-06-29 09:12:20 +00:00

deep_gemm_utils.py

Enable DeepSeek V4 and GLM-5.1 on SM120 (#43477)

2026-06-22 11:54:14 -07:00

eep_reconfigure.py

[MoE Refactor] FusedMoE/MoERunner inversion refactor (#41184)

2026-06-08 10:42:58 -04:00

expert_map_manager.py

[MoE Refactor] Move expert map related code into ExpertMapManager class (#41046)

2026-05-12 09:18:27 -04:00

fused_flydsl_moe.py

[CI/Build][BugFix][The Rock] Fix get_ssm_device_name to return sanitized, usable filename (#47781)

2026-07-08 12:12:54 +00:00

fused_moe_method_base.py

[MoE Refactor] FusedMoE/MoERunner inversion refactor (#41184)

2026-06-08 10:42:58 -04:00

fused_moe_modular_method.py

[MoE Refactor] FusedMoE/MoERunner inversion refactor (#41184)

2026-06-08 10:42:58 -04:00

fused_moe.py

[CI/Build][BugFix][The Rock] Fix get_ssm_device_name to return sanitized, usable filename (#47781)

2026-07-08 12:12:54 +00:00

hpc_moe.py

[MoE Backend] add HPC-Ops MoE backend (#45924)

2026-06-27 11:18:07 +08:00

layer.py

[docs] Fix the docs build (#48008)

2026-07-08 15:47:22 -07:00

modular_kernel.py

[Optimization] Skip DP padding tokens in MoE (#46428)

2026-06-23 14:54:46 -07:00

moe_align_block_size.py

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

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

moe_fused_mul_sum.py

[Quantization] add humming quantization kernel (#34556)

2026-04-24 09:29:44 -04:00

moe_permute_unpermute.py

[Bug] Fix torch device issue for MOE permute (#44005)

2026-05-29 18:55:00 +00:00

routed_experts_capturer.py

[RFC] Replace shared-memory routed experts with ModelRunnerOutput transfer and HTTP support (#39568)

2026-05-14 14:12:30 +00:00

routed_experts.py

[ROCm][CI][MoE] Fix double-transpose of fused w3 expert weights (#47874)

2026-07-08 16:59:28 -07:00

topk_weight_and_reduce.py

Eliminate redundant MoE buffer copies in AITER fused experts (without dependency on AITER changes) (#41713)

2026-05-07 03:46:40 -07:00

unquantized_fused_moe_method.py

[MoE] [MoE Refactor] Add moe kernel oracle abc 37753 (#43461)

2026-06-25 09:34:03 -04:00

utils.py

[MoE] FI autotuning: max bucket = max token count [e.g. DP_size*MNBT] (#47427)

2026-07-07 12:08:36 +03:00