Skip to content

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

New pull request New
44 Open 4,483 Closed
44 Open 4,483 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

Use SPV_INTEL_rounded_divide_sqrt for precise math lowering
#7136 opened Jun 6, 2026 by whitneywhtsang Contributor • Draft
1
iso and unskip all
#7134 opened Jun 5, 2026 by quinnlp Contributor • Draft
[XPU] Allow rematerialization of single-use expensive loads in RemoveLayoutConversions run-benchmarks
#7133 opened Jun 5, 2026 by etiotto Contributor • Draft
1
@etiotto
[TritonIntelGPU] Anchor transpose convert feeding descriptor store run-benchmarks
#7130 opened Jun 5, 2026 by etiotto Contributor Loading…
1
@etiotto
1
Update PyTorch pin
#7127 opened Jun 5, 2026 by anmyachev Contributor • Draft
1
[OptimizeDotOperands] Fix crash on rank-reducing descriptor loads
#7123 opened Jun 5, 2026 by whitneywhtsang Contributor Loading…
2
[ElementwiseOpToLLVM] Lower math.sin/cos to native HW under TRITON_INTEL_FAST_MATH=1
#7120 opened Jun 4, 2026 by etiotto Contributor Loading…
1
@etiotto
4
Fix issue in TDESC load/store lowering for reduced rank.
#7114 opened Jun 4, 2026 by chengjunlu Contributor Loading…
7
[FuseReshape] Gate on ttig.block_io attribute presence
#7112 opened Jun 4, 2026 by whitneywhtsang Contributor • Draft
1
4
[vLLM] Allow overriding triton-repo-branch in vLLM workflows and remove implicit vllm installation from test paths in test-triton.sh
#7109 opened Jun 3, 2026 by quinnlp Contributor • Draft
5
[VLLM][BENCHMARKS][2/3] SYCL-TLA provider for fused moe
#7105 opened Jun 3, 2026 by dev-tomek Contributor • Draft
1
2
[VLLM][BENCHMARKS][1/3] SYCL-TLA provider for unified attention
#7088 opened Jun 3, 2026 by dev-tomek Contributor • Draft
1
Swap xpu tensors with cpu in topk ref values
#7037 opened May 28, 2026 by wdziurdz Contributor • Draft
Add Unified Attention Tile_Size Autotuning
#7034 opened May 27, 2026 by sarah12121212 Contributor Loading…
13
Quinnlp/xpu kernels pin
#7033 opened May 27, 2026 by quinnlp Contributor • Draft
Replace FP8E4M3FN→FP16 table-lookup converter with single fmul run-benchmarks
#7029 opened May 27, 2026 by etiotto Contributor Loading…
1
@etiotto
6
Remove test_sort from skiplist
#7027 opened May 27, 2026 by wdziurdz Contributor • Draft
[Benchmarks] Replace benchmark_config_templates with per-file @register decorator run-benchmarks
#7026 opened May 27, 2026 by mieshkiwrk Contributor Loading…
@mieshkiwrk
Fix SYCL-TLA Flash Attention kernel hang: use no-event kernel launch in non-profiling mode
#7023 opened May 27, 2026 by Copilot AI • Draft
@chengjunlu
[BACKEND] Add reinterpret cast support in ConvertLayoutOp lowering.
#7021 opened May 27, 2026 by chengjunlu Contributor Loading…
1
2
Add FP8 benchmark path to vLLM fused MoE benchmark
#6998 opened May 25, 2026 by Copilot AI • Draft
1
@chengjunlu
[DEBUG] Update ConvertLayoutOpConversion: all store ops before load ops run-benchmarks
#6975 opened May 20, 2026 by anmyachev Contributor • Draft
[BlockIO] Enable 2D block load for runtime row strides run-benchmarks
#6974 opened May 20, 2026 by wdziurdz Contributor Loading…
1
12
[Benchmark] Add FP8 input support to flash_attention_benchmark
#6972 opened May 20, 2026 by Copilot AI Loading…
1
@chengjunlu
10
Add TritonXPU grouped GEMM benchmark for vLLM cutlass_grouped_gemm_interface
#6933 opened May 14, 2026 by Copilot AI • Draft
4 tasks done
@chengjunlu
ProTip! Updated in the last three days: updated:>2026-06-03.