Skip to content

Pull requests: flashinfer-ai/flashinfer

New pull request New
175 Open 1,896 Closed
175 Open 1,896 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Fmha] Sparse MLA decode kernel selection heuristics model: dsv3.2 run-ci
#2836 opened Mar 20, 2026 by PerkzZheng Loading…
5 tasks done
4
fix: use correct SMEM capacity for SM120 consumer Blackwell GPUs
#2835 opened Mar 20, 2026 by brandonmmusic-max Loading…
6
feat: bump nvidia-cutlass-dsl to >=4.4.2 run-ci
#2833 opened Mar 20, 2026 by limin2021 Loading…
7 tasks done
17
fix api breaking changes v0.6.7 release blocker label for 0.6.7
#2832 opened Mar 20, 2026 by aleozlx Loading…
5 tasks
4
skip per-pr for draft PRs
#2831 opened Mar 20, 2026 by aleozlx Loading…
5 tasks
4
Add Option For Fixed cta_tile_q
#2830 opened Mar 19, 2026 by frankwang28 Loading…
5 tasks done
1
9
Proposing issue template
#2829 opened Mar 19, 2026 by aleozlx • Draft
5 tasks
5
[Spark unit test] Adjust tolerance for test_xqa, test_logits_processor
#2828 opened Mar 19, 2026 by kahyunnam Loading…
5 tasks done
7
PR auto-labelling
#2827 opened Mar 19, 2026 by aleozlx • Draft
5 tasks
2
fix: Autotuner _find_nearest_profile non-power-of-2 num_tokens, create launchers for all supported tileN in trtllm fused MoE op: moe run-ci
#2821 opened Mar 19, 2026 by amitz-nv Loading…
4 of 5 tasks
14
Support NVFP4 KV for prefill and batch attention kernels
#2820 opened Mar 19, 2026 by Tom-Zheng Loading…
5 tasks done
19
fix: fall back to CUDA events when CUPTI driver version < 13.0
#2818 opened Mar 19, 2026 by sha7doww Loading…
2 tasks done
6
[fix] Fix TRTLLM MOE autotuner token bucket mismatch error op: moe
#2817 opened Mar 19, 2026 by jiahanc • Draft
5 tasks done
8
Planned CODEOWNERS & review/approval approach changes
#2815 opened Mar 18, 2026 by aleozlx • Draft
3 of 5 tasks
4
[feat] Add optimized clusters TopK kernel and DeepSeekV3.2 sparse indexer fused kernel run-ci
#2814 opened Mar 18, 2026 by Aalanli • Draft
35
[do not merge] fmhav2 hang run-ci
#2813 opened Mar 18, 2026 by jimmyzho Loading…
5 tasks
4
feat(gdn): add padding index guard for bf16 decode kernel
#2810 opened Mar 17, 2026 by kaixih Loading…
5 tasks
13
(draft) [feat] FP8 MoE per-channel quant support op: moe
#2809 opened Mar 17, 2026 by raayandhar • Draft
4 of 5 tasks
6
Fix bmm_fp8 cublasLt handle usage in autotuned cublas runner #26381 run-ci
#2808 opened Mar 17, 2026 by baonudesifeizhai Loading…
3 of 5 tasks
12
Build mnnvl_moe_alltoall with logger and stringUtils op: comm run-ci
#2807 opened Mar 17, 2026 by tiran Loading…
2 of 4 tasks
7
refactor(docker): consolidate Dockerfiles into one multi-stage file
#2806 opened Mar 17, 2026 by dierksen Loading…
3 tasks
8
[CuTe DSL] Add modular FMHA prefill attention kernel
#2805 opened Mar 17, 2026 by pgera • Draft
1 of 3 tasks
9
Refactor the routing part op: moe
#2803 opened Mar 17, 2026 by ChristinaZ Loading…
5 tasks
13
[fix] bugfix 1419: Add batch size shape validation in decode and prefill run() APIs
#2801 opened Mar 17, 2026 by qsang-nv Loading…
5 tasks
1
17
[RFC]MXFP8 autotune IMA at some batch size op: moe
#2800 opened Mar 17, 2026 by charlotte12l Loading…
5 tasks
7
ProTip! no:milestone will show everything without a milestone.