Issues

• ValueError:"Tensors must be contiguous" on single rank

  module: distributedtest: ut
  intel/torch-xpu-ops#5588 · Filter by author madhumitha0102 opened on Oct 1, 2026 · PT2.15
  1

  

• [release/2.15] test_dynamic_shapes.py::DynamicShapesMiscTests::test_module_deepcopy_dynamic_shapes - AssertionError: Scalars are not equal!

  intel/torch-xpu-ops#5586 · Filter by author mazurjax opened on Oct 1, 2026

• [release/2.15] test_quantized_op_xpu.py::TestQuantizedOpsXPU::test_qthreshold_xpu - AssertionError: Quantized tensor-likes are not close!

  intel/torch-xpu-ops#5585 · Filter by author mazurjax opened on Oct 1, 2026

• [release/2.15] test_legacy_vmap_xpu.py - ImportError: cannot import name 'TestVmapBatchedGradientLegacy' from 'test_legacy_vmap'

  intel/torch-xpu-ops#5583 · Filter by author mazurjax opened on Oct 1, 2026

• [2026-10-01] Failing test cases in CI not related to PR changes

  skipped
  Bug

  intel/torch-xpu-ops#5578 · Filter by author BBBela opened on Oct 1, 2026 · UT Backlog
  2

  

• [xpu-alignment] [26-09-30] _foreach_copy_ on XPU raises NotImplementedError for uint16/uint32/uint64 while CPU and copy_ accept them

  ai_generated
  intel/torch-xpu-ops#5576 · Filter by author torchxpubot opened on Oct 1, 2026

• [xpu-alignment] [26-09-30] XPU isin raises "view size is not compatible" for a non-contiguous test_elements on the brute-force path

  ai_generated
  intel/torch-xpu-ops#5575 · Filter by author torchxpubot opened on Oct 1, 2026

• [xpu-alignment] [26-09-30] XPU eager nextafter returns +0.0 instead of -0.0 when stepping up from the smallest negative subnormal

  ai_generated
  intel/torch-xpu-ops#5574 · Filter by author torchxpubot opened on Oct 1, 2026

• Allow upstream tests to execute in float64 dtype after driver issue has been resolved.

  ut_upstream
  Task

  intel/torch-xpu-ops#5573 · Filter by author BBBela opened on Sep 30, 2026
  2

  

• [Bug Skip]: histc_out unexpected success in PR #5554

  skipped
  intel/torch-xpu-ops#5570 · Filter by author AKloniecki opened on Sep 30, 2026 · UT Backlog
  2

  

• [Release/2.15][PVC] E2E eager performance drops ~ 10%-40% vs release/2.14

  hw: PVCperformanceregression
  intel/torch-xpu-ops#5568 · Filter by author kaileiyx opened on Sep 30, 2026

• [upstream_ut] test_nn.py test_upsamplingNearest2d_launch_fail unexpectedly passes on XPU

  intel/torch-xpu-ops#5567 · Filter by author daisyden opened on Sep 30, 2026

• [upstream_ut] test_nn.py test_spatial_softmax_backward_64bit_indexing exceeds XPU work-group limit

  intel/torch-xpu-ops#5566 · Filter by author daisyden opened on Sep 30, 2026

• [upstream_ut] test_nn.py test_nll_loss_1d_input_backward crashes XPU with GPU segfault

  intel/torch-xpu-ops#5565 · Filter by author daisyden opened on Sep 30, 2026

• [upstream_ut] test_nn.py batchnorm mixed bfloat16 training accuracy mismatch on XPU

  intel/torch-xpu-ops#5564 · Filter by author daisyden opened on Sep 30, 2026
  

• [Release/2.15][PVC] E2E inductor performance drops ~ 10%-90% vs release/2.14

  hw: PVCperformanceregression
  intel/torch-xpu-ops#5563 · Filter by author kaileiyx opened on Sep 30, 2026

• [xpu-alignment] [26-09-29] torch.unique(dim=...) on XPU fails to deduplicate and sort sub-tensors when the input contains NaN

  ai_generated
  intel/torch-xpu-ops#5562 · Filter by author torchxpubot opened on Sep 30, 2026

• abs_xpu not implemented for Float8_e5m2

  test: ut
  intel/torch-xpu-ops#5559 · Filter by author daisyden opened on Sep 29, 2026
  

• torch.xpu.device_count() before lazy init ignores ONEAPI_DEVICE_SELECTOR, so is_available() flips across initialization

  os: Linux
  Bug

  intel/torch-xpu-ops#5555 · Filter by author BBBela opened on Sep 29, 2026
  

• [PyTorch CI] Track XPU accuracy skips in common_methods_invocations.py (moved from torch-xpu-ops#2257)

  pytorch-ci-failure
  intel/torch-xpu-ops#5552 · Filter by author daisyden opened on Sep 29, 2026

• [bot] @torchxpubot fix access requests

  intel/torch-xpu-ops#5547 · Filter by author ZhaoqiongZ opened on Sep 28, 2026

• [Release/2.15] 16 E2E performance drops ~ 20% vs release/2.14

  hw: BMGperformanceregression
  intel/torch-xpu-ops#5544 · Filter by author kaileiyx opened on Sep 28, 2026
  

• [Release/2.15] 3 E2E inductor performance drops ~ 15%-30% vs release/2.14

  hw: BMGperformanceregression
  intel/torch-xpu-ops#5543 · Filter by author kaileiyx opened on Sep 28, 2026
  

• grid_sample: implement mode='bicubic' with 5-D input on XPU

  ai_generatedenhancementmodule: ops
  intel/torch-xpu-ops#5542 · Filter by author torchxpubot opened on Sep 28, 2026
  

• [Release/2.15] 15 E2E eager performance drops ~ 30%-70% vs release/2.14

  hw: BMGhw: PVCperformanceregression
  intel/torch-xpu-ops#5541 · Filter by author kaileiyx opened on Sep 28, 2026
  3