Skip to content

Issues: intel/torch-xpu-ops

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

161 Open 176 Closed
161 Open 176 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
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Clarify branch policy of torch-xpu-ops repo - what's viable/strict branch?
#1025 opened Oct 29, 2024 by dvrogozh
@EikanWang
APIs requiring xpu backend support for UT upstream
#1023 opened Oct 29, 2024 by daisyden
@riverliuintel
sort bool type has issue
#1022 opened Oct 28, 2024 by guizili0
1
@xytintel
Performance: Host overhead: Severe host overhead in sycl::get_kernel_bundle.
#1016 opened Oct 23, 2024 by fengyuan14
@majing921201 @fengyuan14
[PT2.6][Windows] import torch not assert in windows, if install torch XPU on a host without driver installed Windows
#1013 opened Oct 23, 2024 by riverliuintel PT2.6 - Feature Freeze
@ratnampa
Logcumsumexp has different results between CPU and XPU on BF16/Complex64/Complex128
#1012 opened Oct 23, 2024 by LuFinch
1
KLDivLoss function in pytorch always fallsback to CPU
#1011 opened Oct 23, 2024 by jgtong
[PT2.6][UX]Torch initialize and verbose message are clear and comparable with PyTorch GPUs enhancement New feature or request feature
#1009 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
[PT2.6] Single AOT binary for at least 3 HWs target with comparable binary size of PyTorch CUDA. feature
#1008 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@fengyuan14
[PT2.6] Pip install Torch and all PyPI dependency feature
#1007 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@chuanqi129
[PT2.6] XPU CI/CD enhancement feature
#1006 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@chuanqi129
[PT2.6] LLM customized kernels (GEMM INT4 serves for Torchao) feature
#1005 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@ZhiweiYan-96
[PT2.6][Torch.complie] Performance analysis and optimization feature performance
#1004 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@retonym @jianyizh @weishi-deng
[PT2.6] INT8 quantization (PT2E) feature on Linux feature
#1003 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@ZhiweiYan-96
[PT2.6] [Torch.complie] Implement AOTInuctor and torch.export on Intel GPU feature
#1002 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@etaf
[PT2.6] Implement >= 80% Aten ops support feature
#1001 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@xytintel
[PT2.6] Front-end support of torchtune, torchao, and torchvision feature
#1000 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@Stonepia
[PT2.6][Client GPU Windows] Windows build log is too big to open Windows
#999 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
@min-jean-cho
1
[PT2.6][Client GPU Windows] Reach comparable three benchmarks pass rate with PVC on LNL Windows Windows
#998 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
[PT2.6][Client GPU Windows] Reach comparable UT pass rate with PVC on LNL Windows Windows
#997 opened Oct 22, 2024 by riverliuintel PT2.6 - Feature Freeze
[Build] PyTorch build failed with new oneAPI and WERROR=1
#987 opened Oct 21, 2024 by mengfei25
1
c10d::allgather_ and c10d::allreduce_ is not currently implemented for the XPU device
#986 opened Oct 19, 2024 by uniartisan
Investigating CompositeExplicitAugograd dispatch key feature
#982 opened Oct 18, 2024 by xytintel
1
[E2E Accuracy] timm jx_nest_base amp_fp16 inference accuracy failed randomly Accuracy amp_fp16 E2E inference timm
#979 opened Oct 17, 2024 by mengfei25 PT2.6
Performance: Reduction: Worse host overhead compared with IPEX performance
#970 opened Oct 16, 2024 by fengyuan14 PT2.6
@majing921201 @fengyuan14
1
ProTip! Mix and match filters to narrow down what you’re looking for.