Commits

Branch selector

 main

User selector

SageMoore

Datepicker

All time

Commit History

Commits on Nov 11, 2025

• [EPLB] Refactor balance_packing to use numpy and optimize GPU-CPU transfers in EPLB (#28369)

  Show description for 798c7be
  SageMoore

  authored
  798c7be

  Copy full SHA for 798c7be

Commits on Nov 10, 2025

• [Bugfix][EPLB] Disabled shared expert overlap when EPLB is enabled (#28377)

  Show description for 40d3326
  

  3 peopleauthored
  40d3326

  Copy full SHA for 40d3326

Commits on Oct 10, 2025

• [Bugfix] Make DP padding optional in coordinate_batch_across_dp (#26375)

  Show description for ae9d0e7
  SageMoore

  authored
  ae9d0e7

  Copy full SHA for ae9d0e7

Commits on Oct 9, 2025

• [BUGFIX] Add cu_tokens_across_sp to DPMetadata (#26457)

  Show description for 0d37450
  SageMoore

  authored
  0d37450

  Copy full SHA for 0d37450

Commits on Oct 7, 2025

• [Core] Simplify the Dp padding/should ubatch coordination logic (#25768)

  Show description for 2111b46
  

  SageMoore

  and

  mgoin

  authored
  2111b46

  Copy full SHA for 2111b46

• [Docs][DBO] Add initial doc that describes the DBO implementation (#26024)

  Show description for c50901f
  SageMoore

  authored
  c50901f

  Copy full SHA for c50901f

Commits on Oct 3, 2025

• Quick fix for IMA with the Prefix Prefill kernel during graph capture (#25983)

  Show description for 5f2cacd
  SageMoore

  authored
  5f2cacd

  Copy full SHA for 5f2cacd

Commits on Sep 26, 2025

• Reduce the Cuda Graph memory footprint when running with DBO (#25779)

  Show description for 4778b42
  SageMoore

  authored
  4778b42

  Copy full SHA for 4778b42

• [Bugfix] Fix Shared Expert/Zero expert code in FusedMoE.process_chunk (#25698)

  Show description for dfb9af2
  

  SageMoore

  and

  robertgshaw2-redhat

  authored
  dfb9af2

  Copy full SHA for dfb9af2

Commits on Sep 24, 2025

• Suppress benign cuBLAS warning when capturing cudagraphs with DBO (#25596)

  Show description for f84a472
  SageMoore

  authored
  f84a472

  Copy full SHA for f84a472

Commits on Sep 16, 2025

• [Core/DBO][1/N] Add Dual-Batch Overlap mechanism to VLLM (#23693)

  Show description for 5679399
  

  4 peopleauthored
  5679399

  Copy full SHA for 5679399

Commits on Sep 15, 2025

• Update num_tokens_across_dp to use nccl instead of gloo (#24105)

  Show description for 49bfc53
  SageMoore

  authored
  49bfc53

  Copy full SHA for 49bfc53

Commits on Aug 2, 2025

• [Attention][DBO] Add support for "splitting" the CommonAttentionMetadata (#21153)

  Show description for 0edaf75
  SageMoore

  authored
  0edaf75

  Copy full SHA for 0edaf75

Commits on May 1, 2025

• [torch.compile] Add torch inductor pass for fusing silu_and_mul with subsequent scaled_fp8_quant operations (#10867)

  Show description for 460a2b1
  SageMoore

  authored
  460a2b1

  Copy full SHA for 460a2b1

Commits on Apr 16, 2025

• [ROCM] Bind triton version to 3.2 in requirements-built.txt (#16664)

  Show description for 44fa4d5
  SageMoore

  authored
  44fa4d5

  Copy full SHA for 44fa4d5

Commits on Mar 21, 2025

• [ROCM] Upgrade torch to 2.6 (#15244)

  Show description for 2e0b4cf
  SageMoore

  authored
  2e0b4cf

  Copy full SHA for 2e0b4cf

Commits on Mar 12, 2025

• [ROCm] Enable chunked prefill/paged attention in MLA on ROCm (#14316)

  Show description for d9f83d6
  SageMoore

  authored
  d9f83d6

  Copy full SHA for d9f83d6

• [ROCm][Bugfix] Ensure that the moe_wna16_gemm kernel is not built on ROCm platforms. (#14629)

  Show description for 45f3f3f
  SageMoore

  authored
  45f3f3f

  Copy full SHA for 45f3f3f

Commits on Mar 4, 2025

• [ROCm] Disable a few more kernel tests that are broken on ROCm (#14145)

  Show description for 3e1d223
  SageMoore

  authored
  3e1d223

  Copy full SHA for 3e1d223

Commits on Mar 1, 2025

• [ROCm][V1][Bugfix] Add get_builder_cls method to the ROCmAttentionBackend class (#14065)

  Show description for b28246f
  SageMoore

  authored
  b28246f

  Copy full SHA for b28246f

Commits on Feb 27, 2025

• [ROCm] Fix the Kernels, Core, and Prefix Caching AMD CI groups (#13970)

  Show description for 38acae6
  SageMoore

  authored
  38acae6

  Copy full SHA for 38acae6

• [ROCm][V1] Update reshape_and_cache to properly work with CUDA graph padding (#13922)

  SageMoore

  authored
  378b3ef

  Copy full SHA for 378b3ef

Commits on Feb 26, 2025

• [ROCm] Disable chunked prefill/prefix caching when running MLA on non-cuda platforms (#13844)

  Show description for 1d35662
  SageMoore

  authored
  1d35662

  Copy full SHA for 1d35662

Commits on Feb 22, 2025

• [V1][Kernel] Refactor the prefix_prefill kernel so that the caller no longer has to pass in the context lengths (#13095)

  SageMoore

  authored
  558db80

  Copy full SHA for 558db80

Commits on Feb 15, 2025

• [Bugfix][AMD] Update torch_bindings so that scaled_fp4_quant isn't build on ROCm (#13235)

  SageMoore

  authored
  c9f9d5b

  Copy full SHA for c9f9d5b

Commits on Feb 14, 2025

• [ROCm][V1] Add intial ROCm support to V1 (#12790)

  SageMoore

  authored
  ba59b78

  Copy full SHA for ba59b78

Commits on Nov 26, 2024

• custom allreduce + torch.compile (#10121)

  Show description for 9a88f89
  

  SageMoore

  and

  youkaichao

  authored
  9a88f89

  Copy full SHA for 9a88f89

Commits on Aug 1, 2024

• [CI/Build] Update PyTorch to 2.4.0 (#6951)

  Show description for 7e0861b
  

  SageMoore

  and

  mgoin

  authored
  7e0861b

  Copy full SHA for 7e0861b

Commits on Mar 2, 2024

• Add Automatic Prefix Caching (#2762)

  Show description for ce4f5a2
  

  3 peopleauthored
  ce4f5a2

  Copy full SHA for ce4f5a2