3 people 6849ca696a [Quantization] Add ModelOpt IQ2_XS, IQ2_XXS, and Q8_0 support via b12x (#59604) ...

Signed-off-by: Andrii Skliar <askliar@nvidia.com>
Signed-off-by: Luke Alonso <lalonso@gmail.com>
Co-authored-by: Andrii Skliar <askliar@nvidia.com>
Co-authored-by: Luke Alonso <lalonso@gmail.com>

2026-10-13 09:06:33 -07:00

..

modular_kernel_tools

[ROCm][MoE] Split AiterExperts by weight-quantization key (#60322)

2026-10-13 00:13:33 -05:00

__init__.py

[Kernel] DeepEP dispatch-combine kernel integration (#18434)

2025-06-03 12:30:02 -07:00

conftest.py

[Bugfix][MoE] Reject hash routing for unsupported monolithic backends (#57867)

2026-09-21 16:28:35 -04:00

parallel_utils.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_b12x.py

[Quantization] Add ModelOpt IQ2_XS, IQ2_XXS, and Q8_0 support via b12x (#59604)

2026-10-13 09:06:33 -07:00

test_batched_deepgemm.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_batched_moe.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_block_fp8.py

[Perf][MoE] Skip top-k slots routed to non-local experts in TritonExp… (#58051)

2026-09-23 21:03:21 +08:00

test_block_int8.py

[ROCm] Fix CI runtime and tests for MI355 DPX (#58244)

2026-09-24 23:15:11 -05:00

test_count_expert_num_tokens.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_cpu_fused_moe.py

[HARDWARE][POWER] Enable W8A8 INT8 MoE on POWER (#55316)

2026-09-17 21:45:13 +08:00

test_cpu_int4_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_cpu_quant_fused_moe.py

[CPU] Add CPU FP8 W8A8 linear/MoE support (#49942)

2026-09-18 17:59:42 +08:00

test_cutedsl_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_cutlass_moe.py

[Bugfix][MoE] Apply expert_map remapping for CUTLASS FP4 MOE expert parallel (#59042)

2026-10-12 10:02:56 -04:00

test_deepep_deepgemm_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_deepep_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_deepep_v2_async_finalize.py

[Kernel][MoE] DeepEP v2: async finalize to overlap shared experts with combine (#52781)

2026-09-15 00:04:22 -04:00

test_deepep_v2_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_deepgemm.py

[Bugfix] Widen ep_gather output offset to int64 (#58373)

2026-10-07 01:45:04 +00:00

test_expert_substitution.py

Add MoNe experts pruning support (#50415)

2026-10-12 23:18:43 +00:00

test_flashinfer_b12x_moe.py

[BugFix] weights processing peak memory reduction for nvfp4 MoE layers (#46276)

2026-07-11 02:05:35 +00:00

test_flashinfer_cutedsl_layout.py

[Kernel] Support Nvfp4 Cutedsl Moe Swiglu-oai and Relu2(non-gated) Activation (#47106)

2026-08-06 17:42:45 -07:00

test_flashinfer_cutedsl_nvfp4_moe.py

[Quantization] Add per-token NVFP4 CuTe-DSL MoE backend (#50030)

2026-09-30 10:42:03 +00:00

test_flashinfer_moe_ep.py

[Bugfix] Drop enable_in_kernel_fc2_reduce from FlashInfer CuteDSL MegaMoE config (#59900)

2026-10-12 09:05:09 -05:00

test_flashinfer_moe.py

Revert "[Bugfix][MoE] Tune FlashInfer experts to scheduler token limit" (#52989) (#53186)

2026-08-21 12:39:28 -06:00

test_flashinfer.py

[Bugfix] Bind routed-experts capture to the MoE layer, not the kernel (#59455)

2026-10-02 15:50:38 +00:00

test_flydsl_moe.py

[ROCm][CI] Stabilize MI355 FlyDSL MoE accuracy test (#53024)

2026-08-21 13:17:18 -07:00

test_fp8_tp_loading.py

[Qwen3.8-Flash-Next] Enable FP8 TP with FlashInfer TRTLLM MoE (#55867)

2026-09-17 08:56:28 +00:00

test_fused_globalize_align_block_size.py

[Kernel] Fuse DeepEP-v2 decode globalize+align into one launch (#53911)

2026-10-13 11:10:38 -04:00

test_fused_topk.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_gemma4router.py

[FEAT] [Perf] [Gemma4] Fused Gemma4 Routing Function Triton (#39083)

2026-04-19 09:57:39 +00:00

test_gpt_oss_triton_kernels.py

[Kernel] Enable OAI Triton MXFP4 MoE on SM12x (#58877)

2026-10-06 01:49:13 -04:00

test_grouped_topk.py

[MoE][Bugfix] Skip SP padded rows in grouped MoE routing (#56079)

2026-09-17 19:36:19 +00:00

test_marlin_vs_trtllm_mxint4.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_modular_kernel_combinations.py

[ROCm][MoE] Split AiterExperts by weight-quantization key (#60322)

2026-10-13 00:13:33 -05:00

test_modular_oai_triton_moe.py

[ROCm][CI] Adapt MoE tests to the triton_kernels 3.8 API (#57385)

2026-09-17 11:21:22 -05:00

test_moe_align_block_size.py

[Feature][Humming] Humming feature integration (#56685)

2026-09-20 08:11:01 -07:00

test_moe_fused_mul_sum.py

[Perf] Tune MoE weighted-sum kernel launch configuration (#59731)

2026-10-02 17:09:29 +00:00

test_moe_kernel_oracle.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_moe_layer.py

[Refactor] Remove dead tests utils (#58803)

2026-09-26 09:16:53 -04:00

test_moe_permute_unpermute.py

[Bugfix][MoE] Size the permute scratch from its inputs (#60447)

2026-10-08 15:32:07 -04:00

test_moe_weight_loading_padded.py

[Bugfix] Load stacked expert weights for non-gated MoE (#59031)

2026-10-05 15:26:55 +03:00

test_moe.py

[Bugfix][MoE] Size the permute scratch from its inputs (#60447)

2026-10-08 15:32:07 -04:00

test_moonep_bf16_poc.py

[Distributed][MoonEP] BF16 integration of MoonEP balanced EP backend (#52101)

2026-09-18 10:34:01 -04:00

test_mori_moe.py

[ROCm][MoE] Split AiterExperts by weight-quantization key (#60322)

2026-10-13 00:13:33 -05:00

test_mori_prepare_finalize.py

[ROCm] Upgrade MoRI version on rocm dockers required for WideEP DP16 DI CI enablement and fixes for combine API (#56073)

2026-09-29 18:07:31 +00:00

test_mxfp4_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_mxfp8_aiter_backend_selection.py

[ROCm][Bugfix] Gate AITER MXFP8 MoE on the aiter enable flag (#57426)

2026-09-18 00:06:05 +00:00

test_nvfp4_moe.py

[MoE] Remove inplace fused experts mechanism (#43727)

2026-05-27 20:00:19 -07:00

test_ocp_mx_moe.py

[Bugfix][MoE] Apply expert_map remapping for CUTLASS FP4 MOE expert parallel (#59042)

2026-10-12 10:02:56 -04:00

test_profile_modular_kernel.py

[Bugfix] Fix broken profile_modular_kernel.py (#43300)

2026-06-11 12:17:23 +01:00

test_rocm_aiter_moe.py

[ROCm][MoE] Split AiterExperts by weight-quantization key (#60322)

2026-10-13 00:13:33 -05:00

test_rocm_aiter_num_expert_group.py

[ROCm][MoE] Pad the AITER MoE intermediate size at allocation time, and round the expert-group count to a kernel that exists (#55368)

2026-09-30 23:08:21 +08:00

test_rocm_aiter_topk.py

[MoE] Move various experts classes to fused_moe/experts/ (#41979)

2026-05-11 07:54:33 +08:00

test_routed_experts_capture_monolithic.py

[Bugfix] Bind routed-experts capture to the MoE layer, not the kernel (#59455)

2026-10-02 15:50:38 +00:00

test_routing_simulator.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_routing.py

[Perf][MoE] Use fused MiniMax2 routing with non-unit routed scaling (#58880)

2026-09-27 19:37:33 +00:00

test_shared_fused_moe_routed_transform.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_silu_mul_fp8_quant_deep_gemm.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

test_silu_mul_per_token_group_quant_fp8_colmajor.py

[Perf][MoE] Skip padded work in block-FP8 DeepGEMM experts (#59128)

2026-10-06 18:49:57 -05:00

test_situ_mul_fp8_quant.py

[MoE] Generalize masked activation for padded layouts (#51217)

2026-09-01 11:34:13 -04:00

test_topk_softplus_sqrt.py

[ROCm][MoE] Split AiterExperts by weight-quantization key (#60322)

2026-10-13 00:13:33 -05:00

test_triton_moe_no_act_mul.py

[XPU][Tests] Make tests device-agnostic (#51968)

2026-08-20 02:25:41 -07:00

test_triton_moe_ptpc_fp8.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

test_trtllm_bf16_moe.py

[MoE] Encapsulate TRT-LLM BF16 weight layout handling (#57405)

2026-09-17 17:47:30 -04:00

test_trtllm_nvfp4_moe.py

[quantization] Add per-token NVFP4 quantization MoE support for ReLU2 (#56740)

2026-10-07 22:01:09 +00:00

test_unquantized_backend_selection.py

[Bugfix] Fall back to auto for unquantized MoE layers when --moe-backend is marlin (#60219)

2026-10-13 08:48:13 +00:00

test_zen_cpu_fused_moe.py

[Model] Add DeepSeek-V4 CPU backend (#55355)

2026-09-10 14:14:48 +08:00

test_zen_cpu_int8_moe.py

[CPU][Zen] Pass f32 weight scales to the zentorch INT8 MoE (#59434)

2026-10-01 15:51:53 +08:00

test_zero_expert_moe.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

utils.py

[Refactor] Remove dead tests code (#58916)

2026-09-30 04:39:03 +00:00