Skip to content

Pull requests: tile-ai/TileOPs

New pull request New
12 Open 1,056 Closed
12 Open 1,056 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

[Perf][foundry][Pool] Take the branches and the 64-bit index out of the max-pool window foundry Kernel generated by the TileFoundry tool perf Performance improvements
#2064 opened Sep 5, 2026 by lcy-seso Collaborator Loading…
5
[Perf][Pool] Rework with-indices max-pool argmax and addressing (1D/2D/3D) perf Performance improvements
#2063 opened Sep 4, 2026 by RMLYC Collaborator Loading…
1
[Feat][Attention] Support dynamic sequence lengths in dense GQA refactor Code restructuring without behavior change
#2062 opened Sep 4, 2026 by superAngGao Collaborator • Draft
[Refactor][Attention] Establish Varlen and Paged GQA Op boundaries refactor Code restructuring without behavior change
#2061 opened Sep 4, 2026 by superAngGao Collaborator • Draft
[Perf][foundry][Norm] Give a norm row a reduction and a width the shape can use foundry Kernel generated by the TileFoundry tool perf Performance improvements
#2060 opened Sep 4, 2026 by lcy-seso Collaborator Loading…
1
[Perf][foundry][Convolution] Fold the Conv2d staging, search rasterization foundry Kernel generated by the TileFoundry tool perf Performance improvements
#2057 opened Sep 3, 2026 by lcy-seso Collaborator Loading…
2
[Feat][Attention] Support RoPE in dense GQA feature New feature or new operator
#2052 opened Sep 3, 2026 by superAngGao Collaborator Loading…
22
[Feat][GEMM] Add the SM90 FP8 1D2D block-scaled GEMM kernel perf Performance improvements
#2040 opened Sep 2, 2026 by michaelwithu Collaborator • Draft
[Bench][Roofline] Judge the bytes audit on the read side only bench Benchmark updates
#1996 opened Aug 27, 2026 by lcy-seso Collaborator • Draft
1
[Perf][GLA] Add optimized inference prefill pipeline perf Performance improvements
#1931 opened Aug 17, 2026 by superAngGao Collaborator • Draft
[Refactor][Attention] Consolidate dense GQA forward topology refactor Code restructuring without behavior change
#1926 opened Aug 15, 2026 by superAngGao Collaborator • Draft
1
[Perf][Attention] Optimize FP8 GQA prefill pipeline
#1873 opened Aug 10, 2026 by superAngGao Collaborator • Draft
ProTip! Add no:assignee to see everything that’s not assigned.