Commits

Branch selector

 main

User selector

elvischenv

All time

Commit History

Commits on Feb 11, 2026

• [GPT-OSS] Remove unnecessary contiguous (#34337)

  Show description for 83e26c8
  elvischenv

  authored
  83e26c8

  Copy full SHA for 83e26c8

Commits on Jan 21, 2026

• Bump Flashinfer to v0.6.1 (#30993)

  Show description for 808d6fd
  elvischenv

  authored
  808d6fd

  Copy full SHA for 808d6fd

Commits on Dec 3, 2025

• [Bugfix] Fix flashinfer ar+norm kernel not available issue (#29960)

  Show description for afe9eb4
  elvischenv

  authored
  afe9eb4

  Copy full SHA for afe9eb4

• [Bugfix] Defunctionalize TRTLLM AR+Norm op for avoiding extra clone kernel before it (#29631)

  Show description for c719c40
  

  elvischenv

  and

  ProExpertProg

  authored
  c719c40

  Copy full SHA for c719c40

Commits on Nov 25, 2025

• [Bugfix] Fix GPT-OSS AR+NORM fusion (#28841)

  Show description for 6330f94
  elvischenv

  authored
  6330f94

  Copy full SHA for 6330f94

Commits on Nov 13, 2025

• [Perf] Support stream interval for reducing host overhead (#27869)

  Show description for 5d6ce2b
  

  elvischenv

  and

  njhill

  authored
  5d6ce2b

  Copy full SHA for 5d6ce2b

Commits on Oct 9, 2025

• [Flashinfer][gpt-oss] Support FP8-qkv Flashinfer TRTLLM Sinks Attention (#25674)

  Show description for 44f633d
  elvischenv

  authored
  44f633d

  Copy full SHA for 44f633d

• Bump Flashinfer to v0.4.0 (#26326)

  Show description for 5e49c3e
  elvischenv

  authored
  5e49c3e

  Copy full SHA for 5e49c3e

Commits on Oct 8, 2025

• [Bugfix][Flashinfer] fix VLLM_USE_TRTLLM_ATTENTION issue for models with diff hyperparameters (#25924)

  Show description for b82f430
  elvischenv

  authored
  b82f430

  Copy full SHA for b82f430

Commits on Sep 17, 2025

• [Bugfix] Fix accuracy issue for silu_mul + nvfp4 quant fusion kernel (#24833)

  Show description for e6585dd
  

  elvischenv

  and

  yewentao256

  authored
  e6585dd

  Copy full SHA for e6585dd

• [Bugfix] Refactor Flashinfer TRTLLM attention kernel selection logic (#24600)

  Show description for e67a79d
  

  elvischenv

  and

  mgoin

  authored
  e67a79d

  Copy full SHA for e67a79d

• [Doc] Add --force-overwrite option to generate_cmake_presets.py (#24375)

  Show description for 3059b9c
  elvischenv

  authored
  3059b9c

  Copy full SHA for 3059b9c

Commits on Sep 13, 2025

• [Perf] Use NVIDIA hardware-accelerated instruction for float to fp8_e4m3 quantization (#24757)

  Show description for dbeee38
  elvischenv

  authored
  dbeee38

  Copy full SHA for dbeee38

Commits on Sep 9, 2025

• [Flashinfer] Support Flashinfer TRTLLM FP8-qkv BF16/FP16-out Attention Kernel (#23647)

  Show description for bba1042
  elvischenv

  authored
  bba1042

  Copy full SHA for bba1042

Commits on Sep 6, 2025

• [Bugfix] Fix unstable silu_mul+nvfp4 quant fusion test (#24370)

  Show description for e68dc2f
  elvischenv

  authored
  e68dc2f

  Copy full SHA for e68dc2f

Commits on Sep 5, 2025

• [Bugfix] Fix silu_mul+quant fusion test (#24341)

  Show description for eedb2a2
  elvischenv

  authored
  eedb2a2

  Copy full SHA for eedb2a2

Commits on Sep 4, 2025

• [Bugfix][Misc] Fix silu_and_mul_nvfp4_quant issue and extract common utils for nvfp4 kernel source files (#23727)

  Show description for adc3ddb
  

  elvischenv

  and

  ProExpertProg

  authored
  adc3ddb

  Copy full SHA for adc3ddb

Commits on Aug 28, 2025

• [NVIDIA] Support SiluMul + NVFP4 quant fusion (#23671)

  Show description for 16a45b3
  

  4 peopleauthored
  16a45b3

  Copy full SHA for 16a45b3

Commits on Aug 22, 2025

• [NVIDIA][torch.compile] Support Flashinfer TRTLLM FP8-q/kv NVFP4-out Attention Kernel (#22703)

  Show description for 24d0c9e
  

  elvischenv

  and

  ProExpertProg

  authored
  24d0c9e

  Copy full SHA for 24d0c9e

Commits on Aug 19, 2025

• [NVIDIA] Support Flashinfer TRTLLM FP8-q/kv/out Attention Kernel (#21716)

  Show description for 03752db
  

  3 peopleauthored
  03752db

  Copy full SHA for 03752db

Commits on Aug 5, 2025

• [NVIDIA] Support Flashinfer TRT-LLM Prefill Attention Kernel (#22095)

  Show description for 83156c7
  elvischenv

  authored
  83156c7

  Copy full SHA for 83156c7

• [Misc] Remove pass_config from CompilationConfig dump_json excluded (#21911)

  Show description for 7b455cf
  elvischenv

  authored
  7b455cf

  Copy full SHA for 7b455cf

Commits on Jul 29, 2025

• [Bugfix] Fix workspace buffer None issue for Flashinfer TRTLLM Backend (#21525)

  Show description for 58b11b2
  elvischenv

  authored
  58b11b2

  Copy full SHA for 58b11b2

Commits on Jul 24, 2025

• [Fix] Update mamba_ssm to 2.2.5 (#21421)

  Show description for 5a19a6c
  elvischenv

  authored
  5a19a6c

  Copy full SHA for 5a19a6c

Commits on Jul 23, 2025

• [Bugfix][CUDA] fixes CUDA FP8 kv cache dtype supported (#21420)

  Show description for 2dec7c1
  elvischenv

  authored
  2dec7c1

  Copy full SHA for 2dec7c1

Commits on Jul 18, 2025

• [Bugfix] Fix the tensor non-contiguous issue for Flashinfer TRT-LLM backend attention kernel (#21133)

  elvischenv

  authored
  8dfb45c

  Copy full SHA for 8dfb45c