Skip to content

Pull requests: intel/intel-xpu-backend-for-triton

New pull request New
51 Open 4,845 Closed
51 Open 4,845 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

[SGLANG] Add own test workflow with PVC/BMG scheduled wrappers run-sglang-tests
#7725 opened Aug 11, 2026 by dev-tomek Contributor • Draft
[LoadStoreOpToLLVM] Re-land eviction_policy -> LSC cache modes without the nontemporal regression
#7724 opened Aug 10, 2026 by etiotto Contributor Loading…
2
@etiotto
7
[CI] Use self-hosted linux runner for check-paths jobs
#7723 opened Aug 10, 2026 by kwasd Contributor Loading…
[CI] Make vLLM tests reusable workflow rolling/lts aware on PVC
#7722 opened Aug 10, 2026 by kwasd Contributor Loading…
1
[TritonIntelGPU] Enable AnnotateCacheControl pass by default except on Windows
#7717 opened Aug 10, 2026 by etiotto Contributor Loading…
1
@etiotto
[Launcher] Add packed-launch fast path (Fix E)
#7704 opened Aug 9, 2026 by NathanVoldman Contributor • Draft
4 of 7 tasks
5
[Launcher] Cache kernel num_args per specialization
#7703 opened Aug 9, 2026 by NathanVoldman Contributor Loading…
5 of 7 tasks
1
3
Speed up JIT kernel launch by binding the device/stream lookups once lts run-benchmarks
#7695 opened Aug 7, 2026 by mieshkiwrk Contributor Loading…
@mieshkiwrk
Merge OpenAI Triton commit d4882a3
#7694 opened Aug 7, 2026 by wdziurdz Contributor Loading…
[RemoveLayoutConversions] Don't charge free converts an SLM round-trip run-benchmarks run-vllm-benchmarks
#7685 opened Aug 6, 2026 by etiotto Contributor Loading…
1
@etiotto
2
[Launcher] Intern data_ptr string in extractPointer
#7680 opened Aug 6, 2026 by NathanVoldman Contributor Loading…
5 of 7 tasks
1
2
[Frontend] Add runtime assertion for signed div/rem with negative dividend run-vllm-tests
#7674 opened Aug 5, 2026 by whitneywhtsang Contributor • Draft
1
[Launcher] Re-land event-less kernel submission lts run-benchmarks run-vllm-benchmarks
#7670 opened Aug 5, 2026 by mieshkiwrk Contributor Loading…
1
@mieshkiwrk
Improve AccelerateMatmul.cpp for the cases of the chained dot operations. run-benchmarks
#7668 opened Aug 5, 2026 by chengjunlu Contributor Loading…
11
Improve the remove layout in handling loop op.
#7634 opened Aug 4, 2026 by chengjunlu Contributor Loading…
1
13
Split Windows build
#7630 opened Aug 3, 2026 by kwasd Contributor Loading…
Update PyTorch pin
#7622 opened Aug 2, 2026 by anmyachev Contributor Loading…
3
Bucket unified attention autotuning by Q and KV length
#7619 opened Jul 31, 2026 by sarah12121212 Contributor • 2/3 • Draft
Honor XPU maxnreg in auto-GRF retry selection
#7618 opened Jul 31, 2026 by Copilot AI • Draft
1
@chengjunlu
3
[NIT] Rebuild the kernel with large GRF size to reduce the register spilling size aggressively. run-benchmarks
#7615 opened Jul 31, 2026 by chengjunlu Contributor Loading…
1
1
[vLLM Update pin run-vllm-tests
#7613 opened Jul 30, 2026 by quinnlp Contributor • Draft
[LAYOUTS] Dispatch dot-operand and Gluon layouts through interfaces
#7609 opened Jul 29, 2026 by mieshkiwrk Contributor Loading…
[TritonIntelGPUToLLVM] Vectorize masked DescriptorLoadOp/DescriptorStoreOp when boundary mask isn't uniform across a group (fixes OOB access) run-benchmarks
#7590 opened Jul 27, 2026 by anmyachev Contributor • Draft
[TritonIntelGPUToLLVM] Vectorize masked LoadOp when mask isn't statically uniform run-benchmarks
#7587 opened Jul 26, 2026 by anmyachev Contributor • Draft
[AxisInfo][StrideInfo] Disable signed division/remainder optimizations run-benchmarks
#7570 opened Jul 23, 2026 by whitneywhtsang Contributor • Draft
1
1
ProTip! Follow long discussions with comments:>50.