Skip to content

Issues: intel/intel-xpu-backend-for-triton

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

135 Open 1,087 Closed
135 Open 1,087 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

undefined reference to `_Z27__spirv_GroupNonUniformIAddiibj' bug Something isn't working upstream: pytorch
#3109 opened Jan 7, 2025 by Stonepia
@victor-eds
'tt.expand_dims' op inferred type(s) are incompatible with return type(s) of operation bug Something isn't working upstream: pytorch
#3107 opened Jan 7, 2025 by Stonepia
1
UploadMXFPOp optimization with inlining vISA performance
#3106 opened Jan 7, 2025 by LiyangLingIntel
[CI] Upgrade CI infra to Agama Rolling 2448.13 (1057.13) ci driver: rolling
#3100 opened Jan 6, 2025 by vlad-penkin 7. CI
1
@pbchekin
5
SPIRV-Runner crash after signature/constant representation change (PR-3043) bug Something isn't working
#3096 opened Jan 6, 2025 by kballeda 0.3 [Triton] Language and Runtime
@kballeda
[Pytorch Upstream] ImportError: libsycl.so.7: cannot open shared object file bug Something isn't working upstream: pytorch
#3095 opened Jan 6, 2025 by etaf 0.1 [PT 2.6 Upstream] TorchInductor
@pbchekin
NaN for FlexAttention with triton xpu bug Something isn't working upstream: pytorch
#3094 opened Jan 6, 2025 by liangan1 0.1 [PT 2.6 Upstream] TorchInductor
@alexbaden
7
Evaluate new IGC ManageableBarriers research
#3093 opened Jan 6, 2025 by whitneywhtsang 0.3 [Triton] Language and Runtime
@whitneywhtsang
1
Merge OpenAI Triton till Jan 18th upstream: rebase PR to be up-streamed
#3091 opened Jan 5, 2025 by whitneywhtsang 0.2 [Triton] Upstream
@whitneywhtsang
1
[Windows][Pytorch Upstream] The pyd files comes from Triton compile_module_from_src can not be cleaned. bug Something isn't working windows
#3090 opened Jan 4, 2025 by etaf 0.1 [PT 2.6 Upstream] TorchInductor
Merge 67dc627 upstream: rebase PR to be up-streamed
#3087 opened Jan 3, 2025 by whitneywhtsang 0.2 [Triton] Upstream
1
@anmyachev
[Performance] Improve register spill size for flash attention kernel when D_HEAD >=128 enhancement New feature or request performance
#3084 opened Jan 3, 2025 by chengjunlu 4.0 [Performance] Core
@chengjunlu
[Windows][Pytorch Upstream] Triton do_bench returns time not greater than 0 on Windows. bug Something isn't working windows
#3072 opened Dec 27, 2024 by etaf 0.1 [PT 2.6 Upstream] TorchInductor
[Windows] level_zero/ze_api.h not found when running UT. bug Something isn't working windows
#3064 opened Dec 24, 2024 by Stonepia 0.1 [PT 2.6 Upstream] TorchInductor
4
Try to avoid duplication among generated assert constants in .llir file codegen: mlir enhancement New feature or request
#3054 opened Dec 20, 2024 by anmyachev 0.3 [Triton] Language and Runtime
Code changes clean after #3010 code quality enhancement New feature or request
#3046 opened Dec 19, 2024 by ESI-SYD 0.3 [Triton] Language and Runtime
@ESI-SYD
Don't use implicitly elapsed_time in autotuner when profiling with PyTorch and PTI enhancement New feature or request performance profiling
#3039 opened Dec 18, 2024 by anmyachev 4.5 [Performance] Profiling
1
@anmyachev
[win] test_debug does not work on windows bug Something isn't working hw: arc windows
#3018 opened Dec 16, 2024 by pbchekin 5.1 [DGPU] Windows builds
[win] Interpreter tests failures bug Something isn't working windows
#3017 opened Dec 16, 2024 by pbchekin 5.1 [DGPU] Windows builds
[win] test_scaled_dot[128-128-64-False-True-True-e5m2-e5m2-4-16-1] failure bug Something isn't working hw: arc windows
#3016 opened Dec 16, 2024 by pbchekin 5.1 [DGPU] Windows builds
[win] test_precise_math[1-tl.math.sqrt_rn(x)-tl.math.sqrt(x.to(tl.float64)).to(tl.float32)] failure bug Something isn't working windows
#3015 opened Dec 16, 2024 by pbchekin 5.1 [DGPU] Windows builds
Add onednn to gemm benchmarks
#3014 opened Dec 16, 2024 by alexbaden 4.3 [Performance] Tracking
1
@alexbaden
1
PyTorch test error on nightly 1213 version on BMG device, time of benchmarker.benchmark < 0. bug Something isn't working
#3013 opened Dec 16, 2024 by PenghuiCheng 5.0 [DGPU] Initial enabling
@kwasd @anmyachev
2
[e2e test] port flash attention from sglang enhancement New feature or request tests: e2e
#3012 opened Dec 16, 2024 by Dewei-Wang-sh 4.2 [Performance] E2E
1
@Dewei-Wang-sh
2
Add tile size abstraction in tensor ptr TTGIR to LLVM lowering code quality enhancement New feature or request
#3008 opened Dec 14, 2024 by alexbaden 0.3 [Triton] Language and Runtime
@alexbaden
ProTip! Adding no:label will show everything without a label.