Skip to content

Pull requests: intel/torch-xpu-ops

New pull request New
146 Open 1,683 Closed
146 Open 1,683 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

[CI] Remove contents write permission in auto merge workflow disable_all Disable all ci test jobs for the PR, just keep basic lint check
#3407 opened Apr 20, 2026 by chuanqi129 Contributor Loading…
6
Update tags.yaml to align with PyTorch's
#3406 opened Apr 20, 2026 by janeyx99 Loading…
4
Add variable-length support for Flash Attention in XPU
#3405 opened Apr 20, 2026 by tszulist-hbn Contributor • Draft
1
Enable Op _flash_attention_forward/_flash_attention_backward
#3404 opened Apr 20, 2026 by LuFinch Contributor Loading…
1
22
Skip test_fake_crossref_backward_amp_nn_functional_bilinear_xpu_float32
#3403 opened Apr 20, 2026 by jkosnox • Draft
[#3362] Update test_cudnn_weight_format for xpu disable_accelerate Disable accelerate test job in PR CI testing disable_distributed Disable distributed UT test jobs for the PR disable_e2e Disable all e2e test jobs for the PR disable_transformers Disable transformers UT test in PR CI disable_win Disable Windows CI test jobs for the PR
#3402 opened Apr 20, 2026 by jmamzax Contributor Loading…
3
[#2630] Override test_fmod_remainder_by_zero_integral to work with int64 on xpu disable_accelerate Disable accelerate test job in PR CI testing disable_distributed Disable distributed UT test jobs for the PR disable_e2e Disable all e2e test jobs for the PR disable_transformers Disable transformers UT test in PR CI disable_win Disable Windows CI test jobs for the PR
#3401 opened Apr 20, 2026 by jmamzax Contributor Loading…
3
[Perf] Add vectorized scatter kernel for bf16/fp16
#3400 opened Apr 20, 2026 by yanqiuz Loading…
2
test: transformer_encoder_layer: allow xpu for fast_path_device
#3399 opened Apr 20, 2026 by BartoszKokoszko Contributor Loading…
1
Add .flake8 lint configuration with test/xpu paths
#3398 opened Apr 20, 2026 by daisyden Contributor Loading…
5
[#2811] Add long test to skip list to run it only on nightly disable_all Disable all ci test jobs for the PR, just keep basic lint check
#3397 opened Apr 20, 2026 by jmamzax Contributor Loading…
4
[Copilot] Try to fix the permission error with the user id
#3392 opened Apr 20, 2026 by Stonepia Contributor • Draft
XPU: Port dynamo test cases to Intel GPU
#3383 opened Apr 18, 2026 by daisyden Contributor Loading…
22
Rewrite nonzero kernel to use bounded scratch memory disable_accelerate Disable accelerate test job in PR CI testing disable_distributed Disable distributed UT test jobs for the PR disable_e2e Disable all e2e test jobs for the PR disable_transformers Disable transformers UT test in PR CI disable_win Disable Windows CI test jobs for the PR
#3374 opened Apr 17, 2026 by pbielak Contributor • Draft
9
Foreachmax add early exit on empty tensor disable_accelerate Disable accelerate test job in PR CI testing disable_distributed Disable distributed UT test jobs for the PR disable_e2e Disable all e2e test jobs for the PR disable_transformers Disable transformers UT test in PR CI disable_win Disable Windows CI test jobs for the PR
#3373 opened Apr 17, 2026 by Silv3S Contributor Loading…
1
5
Add single workgroup topk kernel for XPU (part 2 of #3369)
#3372 opened Apr 17, 2026 by jianyizh Contributor Loading…
7
Add subgroup topk kernel for XPU (part1 of #3369)
#3371 opened Apr 17, 2026 by jianyizh Contributor Loading…
31
Optimize torch.topk on XPU: sub-group bitonic merge + radix select dispatch
#3369 opened Apr 17, 2026 by jianyizh Contributor • Draft
8
[RELEASE 2.12] Release only changes
#3360 opened Apr 16, 2026 by chuanqi129 Contributor Loading…
1
Fix native_batch_norm eval mode returning populated save_mean/save_invstd on XPU
#3351 opened Apr 16, 2026 by Copilot AI • Draft
1
@Stonepia
13
Preserves the index data type of sparse CSR tensor
#3341 opened Apr 15, 2026 by chunhuanMeng Contributor Loading…
1
3
Improve long float32 torch.cumprod numerical parity on XPU
#3339 opened Apr 15, 2026 by Copilot AI Loading…
2 tasks done
1
@Stonepia
14
Work around incorrect XPU Inductor backward for slice_scatter and add regression coverage
#3338 opened Apr 15, 2026 by Copilot AI • Draft
1
@Stonepia
1
Prevent float32 torch.std/torch.var overflow on XPU for large-magnitude inputs
#3337 opened Apr 15, 2026 by Copilot AI • Draft
1
@Stonepia
11
Fix nondeterministic alert for _histc_xpu: only alert for floating point input
#3333 opened Apr 15, 2026 by chunhuanMeng Contributor Loading…
7
ProTip! Updated in the last three days: updated:>2026-04-17.