Improve conv2d backward pass performance #183512

Dr. CI
Commit:

Commit CI Status

O
Lint — success (1m 12s)
 O
pull / linux-jammy-py3.8-gcc11 / build — success (8m 32s)
 O
pull / linux-jammy-py3.8-gcc11 / test — success (22m 14s)
 O
pull / linux-focal-cuda12.1 / build — success (12m 3s)
 X
pull / linux-focal-cuda12.1 / test — failure FAIL test_sparse (SparseTestCUDA) Duration: 45m 12s
 OOOOOOO??OOO?OOO~
ⓘ This is the old dropdown design. View the new workflow-column design →
▾
Cross-Repository CI (L3/L4) (4/5 passed, 1 running)
BackendJobStatusDurationLinks
intel/intel-extension-for-pytorch L3build-and-test success 27sRun Artifacts
intel/intel-extension-for-pytorch L3test-xpu-ops failure 31sRun Artifacts
pytorch-labs/ascend-pytorch L3build-and-test success 35sRun Artifacts
pytorch-labs/ascend-pytorch L3test-npu running –Run
pytorch-labs/cambricon-pytorch L4build-and-test success 32sRun Artifacts
Mockup Pages: Main HUD · CRCR Summary · Per-Repo (PR) · Nightly · Periodic · PR Workflows · PR View (Legacy)