Skip to content

Pull requests: pytorch/helion

New pull request New
170 Open 2,661 Closed
170 Open 2,661 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

[triton][AMD] Expand tunable values for waves_per_eu and matrix_instr_nonkdim CLA Signed This label is managed by the Meta Open Source bot.
#3123 opened Jul 22, 2026 by ethche Contributor Loading…
Remove stale PLDI tutorial link CLA Signed This label is managed by the Meta Open Source bot.
#3122 opened Jul 21, 2026 by jansel Contributor Loading…
[cute] Enable tcgen05 collective MMA for batched CuTe matmul CLA Signed This label is managed by the Meta Open Source bot.
#3121 opened Jul 21, 2026 by ethche Contributor Loading…
[benchmarks] Limit unstable MI350 input sweeps CLA Signed This label is managed by the Meta Open Source bot.
#3120 opened Jul 21, 2026 by ethche Contributor Loading…
[autotuner] Add multi-shape autotuning CLA Signed This label is managed by the Meta Open Source bot.
#3119 opened Jul 21, 2026 by ethche Contributor Loading…
[Pallas] Materialize each benchmark output so a lazy torch.compile baseline can't inflate neighbors CLA Signed This label is managed by the Meta Open Source bot.
#3111 opened Jul 21, 2026 by norx1991 Contributor Loading…
2
[benchmark] Shard the linear-attention benchmark across parallel jobs CLA Signed This label is managed by the Meta Open Source bot.
#3107 opened Jul 20, 2026 by tarinduj Contributor Loading…
3
[examples] Perf-tune the linear-attention kernels vs FLA CLA Signed This label is managed by the Meta Open Source bot.
#3105 opened Jul 20, 2026 by tarinduj Contributor • Draft
5
Fix static_shapes output-shape leak when two shapes emit identical source CLA Signed This label is managed by the Meta Open Source bot.
#3099 opened Jul 20, 2026 by ethche Contributor Loading…
Fix rank mismatch for size-1 slice store patterns CLA Signed This label is managed by the Meta Open Source bot.
#3098 opened Jul 19, 2026 by parsshar-RH Contributor Loading…
4
[pallas] enable a max_extent hint for jagged hl.tile loops CLA Signed This label is managed by the Meta Open Source bot.
#3091 opened Jul 17, 2026 by ethche Contributor • Draft
4
[Pallas] Lower packed and runtime-lane VMEM scalar loads via pad+roll+widen CLA Signed This label is managed by the Meta Open Source bot.
#3087 opened Jul 17, 2026 by thcmbs Collaborator Loading…
5
[Pallas] Support runtime second-minor scalar loads from VMEM for 32-bit dtypes CLA Signed This label is managed by the Meta Open Source bot.
#3086 opened Jul 17, 2026 by thcmbs Collaborator Loading…
[Pallas] Switch the compact-worklist launcher to pl.kernel CLA Signed This label is managed by the Meta Open Source bot.
#3085 opened Jul 17, 2026 by thcmbs Collaborator Loading…
1
[Pallas] Fold tritonbench-bridge results into the tpu dashboard column CLA Signed This label is managed by the Meta Open Source bot.
#3082 opened Jul 16, 2026 by norx1991 Contributor Loading…
[pallas] ordered carry: support multidimensional tiles CLA Signed This label is managed by the Meta Open Source bot.
#3077 opened Jul 16, 2026 by cota Collaborator Loading…
2
[pallas] ordered carry: allow dynamic number of columns CLA Signed This label is managed by the Meta Open Source bot.
#3076 opened Jul 16, 2026 by cota Collaborator Loading…
4
CI: vLLM Helion kernel correctness + compile check on H100/B200 CLA Signed This label is managed by the Meta Open Source bot.
#3074 opened Jul 16, 2026 by yushangdi Contributor • Draft
5
Per token group fp8 quant.py CLA Signed This label is managed by the Meta Open Source bot.
#3073 opened Jul 16, 2026 by yushangdi Contributor • Draft
5
Gate distributed-only compilation on symmetric-memory usage not dist.is_initialized() CLA Signed This label is managed by the Meta Open Source bot.
#3070 opened Jul 15, 2026 by karthickai Contributor Loading…
1
Yushangdi/trusted fused launch fast path CLA Signed This label is managed by the Meta Open Source bot.
#3068 opened Jul 14, 2026 by yushangdi Contributor • Draft
Decide distributed pid_type restrictions based on kernel tracing CLA Signed This label is managed by the Meta Open Source bot.
#3057 opened Jul 14, 2026 by bringlein Collaborator Loading…
1
[Autotuner] detailed search space analysis CLA Signed This label is managed by the Meta Open Source bot.
#3055 opened Jul 14, 2026 by bringlein Collaborator Loading…
2
Add pretuned CuTe flash attention kernel CLA Signed This label is managed by the Meta Open Source bot.
#3048 opened Jul 14, 2026 by yushangdi Contributor • Draft
1
Add pretuned NVFP4 decode GEMV benchmark vs vLLM CUTLASS CLA Signed This label is managed by the Meta Open Source bot.
#3046 opened Jul 13, 2026 by yushangdi Contributor • Draft
3
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.