Commits

Branch selector

 main

User selector

alexm-redhat

Datepicker

All time

Commit History

Commits on Nov 12, 2025

• [MoE][Kernel][Perf] Improve Shared Expert Stream Overlap (#28406)

  Show description for 69d0e90
  alexm-redhat

  authored
  69d0e90

  Copy full SHA for 69d0e90

• [Performance][Hopper] Avoid M dim padding to 4x for most cases (due to cuda graphs paddings) (#28492)

  Show description for f76e85c
  alexm-redhat

  authored
  f76e85c

  Copy full SHA for f76e85c

Commits on Oct 23, 2025

• [Bugfix] Fix dp_chunking enablement logic in FusedMoE layer (#27220)

  Show description for 9ef3d5b
  alexm-redhat

  authored
  9ef3d5b

  Copy full SHA for 9ef3d5b

Commits on Oct 21, 2025

• [Performance] Dual stream execution of "shared_experts" and "selected_experts" inside FusedMoE (#26440)

  Show description for 344a001
  alexm-redhat

  authored
  344a001

  Copy full SHA for 344a001

Commits on Sep 23, 2025

• [Bugfix] [B200] cutlass_mla - ensure kv_split == 1 for batch size > 1 (#25509)

  Show description for 1210e4d
  alexm-redhat

  authored
  1210e4d

  Copy full SHA for 1210e4d

• [Performance] Remove input pads in cutlass_mla and optimize v_proj output handling (#25184)

  Show description for 0b7bed9
  alexm-redhat

  authored
  0b7bed9

  Copy full SHA for 0b7bed9

Commits on Sep 17, 2025

• [Bugfix][B200] Fix `cutlass_mla` hang (#24966)

  Show description for fedb75f
  

  alexm-redhat

  and

  mgoin

  authored
  fedb75f

  Copy full SHA for fedb75f

Commits on Sep 15, 2025

• [Performance] Remove redundant clone() calls in cutlass_mla (#24891)

  alexm-redhat

  authored
  aae725a

  Copy full SHA for aae725a

Commits on Jul 27, 2025

• [Attention] Make CutlassMLA the default backend for SM100 (blackwell) (#21626)

  Show description for 8f605ee
  

  alexm-redhat

  and

  mgoin

  authored
  8f605ee

  Copy full SHA for 8f605ee

Commits on Jul 15, 2025

• SM100 Cutlass MLA decode with unrestricted num_heads (< 128) for DeepSeek TP (#20769)

  Show description for 8cdc371
  alexm-redhat

  authored
  8cdc371

  Copy full SHA for 8cdc371

Commits on Jul 11, 2025

• [Attention] MLA - Flashinfer Ragged Prefill (#20034)

  alexm-redhat

  authored
  5b03235

  Copy full SHA for 5b03235

Commits on Apr 1, 2025

• [V1] TPU - Fix fused MOE (#15834)

  Show description for 7e4e709
  alexm-redhat

  authored
  7e4e709

  Copy full SHA for 7e4e709

Commits on Mar 31, 2025

• [V1] TPU CI - Add basic perf regression test (#15414)

  Show description for 9a2160f
  alexm-redhat

  authored
  9a2160f

  Copy full SHA for 9a2160f

Commits on Mar 28, 2025

• [V1] TPU - Fix the chunked prompt bug (#15713)

  Show description for c3f687a
  alexm-redhat

  authored
  c3f687a

  Copy full SHA for c3f687a

Commits on Mar 26, 2025

• [V1] TPU CI - Fix test_compilation.py (#15570)

  Show description for 9d119a8
  alexm-redhat

  authored
  9d119a8

  Copy full SHA for 9d119a8

• [V1] TPU - Revert to exponential padding by default (#15565)

  Show description for b2e85e2
  alexm-redhat

  authored
  b2e85e2

  Copy full SHA for b2e85e2

Commits on Mar 20, 2025

• [V1] TPU - Tensor parallel MP support (#15059)

  alexm-redhat

  authored
  cfbca8a

  Copy full SHA for cfbca8a

Commits on Mar 18, 2025

• [V1] TPU - CI/CD use smaller model (#15054)

  Show description for 72a8639
  alexm-redhat

  authored
  72a8639

  Copy full SHA for 72a8639

Commits on Mar 17, 2025

• [V1] TPU - Fix CI/CD runner (#14974)

  alexm-redhat

  authored
  18551e8

  Copy full SHA for 18551e8

Commits on Mar 14, 2025

• [V1] TPU - Enable prefix caching by default (#14773)

  alexm-redhat

  authored
  7888e1d

  Copy full SHA for 7888e1d

Commits on Mar 8, 2025

• [V1] TPU - Add tensor parallel support via Ray (#13618)

  Show description for cb8bdfa
  alexm-redhat

  authored
  cb8bdfa

  Copy full SHA for cb8bdfa

Commits on Feb 14, 2025

• [WIP] TPU V1 Support Refactored (#13049)

  alexm-redhat

  authored
  45f90bc

  Copy full SHA for 45f90bc

Commits on Dec 18, 2024

• [V1] VLM - enable processor cache by default (#11305)

  Show description for fdea8ec
  alexm-redhat

  authored
  fdea8ec

  Copy full SHA for fdea8ec

Commits on Dec 13, 2024

• [Misc] Add tokenizer_mode param to benchmark_serving.py (#11174)

  Show description for 238c0d9
  alexm-redhat

  authored
  238c0d9

  Copy full SHA for 238c0d9

Commits on Dec 12, 2024

• [V1] VLM preprocessor hashing (#11020)

  Show description for 4e11683
  

  3 peopleauthored
  4e11683

  Copy full SHA for 4e11683

Commits on Dec 3, 2024

• [V1] VLM - Run the mm_mapper preprocessor in the frontend process (#10640)

  Show description for 3bc94ca
  

  3 peopleauthored
  3bc94ca

  Copy full SHA for 3bc94ca

Commits on Sep 23, 2024

• Add output streaming support to multi-step + async while ensuring RequestOutput obj reuse (#8335)

  alexm-redhat

  authored
  1a2aef3

  Copy full SHA for 1a2aef3

Commits on Sep 18, 2024

• [Core][Bugfix][Perf] Introduce `MQLLMEngine` to avoid `asyncio` OH (#8157)

  Show description for 7c7714d
  

  5 peopleauthored
  7c7714d

  Copy full SHA for 7c7714d

Commits on Sep 13, 2024

• [Bugfix] Fix async log stats (#8417)

  alexm-redhat

  authored
  6821020

  Copy full SHA for 6821020

Commits on Sep 12, 2024

• [Bugfix] multi-step + flashinfer: ensure cuda graph compatible (#8427)

  alexm-redhat

  authored
  0198772

  Copy full SHA for 0198772

Commits on Sep 10, 2024

• [Bugfix] lookahead block table with cuda graph max capture (#8340)

  Show description for 22f3a4b
  alexm-redhat

  authored
  22f3a4b

  Copy full SHA for 22f3a4b

Commits on Sep 8, 2024

• [Bugfix] Fix async postprocessor in case of preemption (#8267)

  alexm-redhat

  authored
  4ef41b8

  Copy full SHA for 4ef41b8

Commits on Sep 3, 2024

• [Core] Optimize Async + Multi-step (#8050)

  alexm-redhat

  authored
  6d646d0

  Copy full SHA for 6d646d0

Commits on Aug 29, 2024

• [Core] Combine async postprocessor and multi-step (#7921)

  alexm-redhat

  authored
  3f60f22

  Copy full SHA for 3f60f22

Commits on Aug 28, 2024

• [Core] Async_output_proc: Add virtual engine support (towards pipeline parallel) (#7911)

  alexm-redhat

  authored
  f508e03

  Copy full SHA for f508e03

Pagination

Previous

Next