Search results

Tensor-likes are not close: test_mxfp8_mxfp4_matmul on PVC with Kobuk

accuracy

driver: rolling

hw: pvc

tests: ut

Status: Open.

Bug

#5428 In intel/intel-xpu-backend-for-triton;

· kwasd opened on Nov 4, 2025

Use precise divide and sqrt builtin

dependencies: igc

Status: Open.

Feature

#5419 In intel/intel-xpu-backend-for-triton;

· whitneywhtsang opened on Nov 3, 2025

triton-kernels are stuck on A770 Linux

hw: arc

tests: ut

Status: Open.

Bug

#5418 In intel/intel-xpu-backend-for-triton;

· kwasd opened on Nov 3, 2025

<code>PYTORCH_OPINFO_SAMPLE_INPUT_INDEX=2 python test/inductor/test_torchinductor_opinfo.py TestInductorOpInfoXPU.test_comprehensive___rmod___xpu_float32</code> is broken

dependencies: igc

tests: torchinductor

upstream: pytorch

Status: Open.

Bug

#5411 In intel/intel-xpu-backend-for-triton;

· anmyachev opened on Oct 30, 2025

· 

1. [PT 2.10 Upstream] TorchInductor

[perf] Tensor descriptors have poor performance when strides are not tl.constexpr

performance

tests: ut

Status: Open.

Task

#5410 In intel/intel-xpu-backend-for-triton;

· Egor-Krivov opened on Oct 30, 2025

· 

4. [Performance] Core

<code>PYTORCH_OPINFO_SAMPLE_INPUT_INDEX=7 python test/inductor/test_torchinductor_opinfo.py TestInductorOpInfoXPU.test_comprehensive__chunk_cat_xpu_float64</code> is broken

dependencies: igc

tests: torchinductor

upstream: pytorch

Status: Open.

Bug

#5408 In intel/intel-xpu-backend-for-triton;

· anmyachev opened on Oct 29, 2025

[PT 2.10] Triton pin update in PyTorch

dependencies: pytorch

tests: torchinductor

upstream: pytorch

Status: Open.

Task

#5407 In intel/intel-xpu-backend-for-triton;

· anmyachev opened on Oct 29, 2025

· 

1. [PT 2.10 Upstream] TorchInductor

Use pytorch train cnn model torch.compile backend compiler failed build.

dependencies: pytorch

tests: torchinductor

upstream: pytorch

Status: Open.

Bug

#5405 In intel/intel-xpu-backend-for-triton;

· case2012 opened on Oct 29, 2025

· 

1. [PT 2.10 Upstream] TorchInductor

[Intel] Add padding support to tensor descriptor

tests: ut

upstream: triton

Status: Open.

Task

#5400 In intel/intel-xpu-backend-for-triton;

· whitneywhtsang opened on Oct 29, 2025

· 

3. [Triton] Language and Runtime

Error during Intel loadBinary: ZE_RESULT_ERROR_INVALID_KERNEL_NAME

dependencies: pytorch

upstream: pytorch

Status: Open.

Bug

#5394 In intel/intel-xpu-backend-for-triton;

· etaf opened on Oct 28, 2025

· 

1. [PT 2.10 Upstream] TorchInductor

Compatibility section of Readme is outdated

documentationImprovements or additions to documentationImprovements or additions to documentation

Status: Open.

Task

#5372 In intel/intel-xpu-backend-for-triton;

· anmyachev opened on Oct 23, 2025

· 

3. [Triton] Language and Runtime

[Helion] Support rank &gt; 2 tensor descriptor load with block IO.

tests: e2e

upstream: pytorch

Status: Open.

Task

#5365 In intel/intel-xpu-backend-for-triton;

· chengjunlu opened on Oct 22, 2025

· 

3. [Triton] Language and Runtime