TorchedHat/pytorch-redhat-ci : nightly

← Back to CRCR Summary GitHub ↗
🔗 Permalink
Pass Rate
85.7%
30/35 jobs
Nightly Runs
7
unique SHAs tested
Failures
2
+ 3 timed out
Rows = nightly CI runs (one per SHA), columns = build & test stages. Click a cell to open the workflow run.
Time SHA Commit
build
test (cpu)
test (inductor)
test (sgpu)
test (mgpu)
Update torch/distributed/_tensor to use DTensor.from_local instead of deprecated distribute_tensor O
build — success Duration: 1h 49m Show log ›
O
test (cpu) — success Duration: 1m 2s Tests: 2 passed, 0 failed Show log ›
O
test (inductor) — success Duration: 1m 7s Tests: 2 passed, 0 failed Show log ›
O
test (sgpu) — success Duration: 1m 38s Tests: 2 passed, 0 failed Show log ›
O
test (mgpu) — success Duration: 48s Tests: 2 passed, 0 failed Show log ›
[inductor] Fix bug in pattern matching with unbacked symints (#189042) O
build — success Duration: 1h 52m Show log ›
O
test (cpu) — success Duration: 58s Show log ›
O
test (inductor) — success Duration: 1m 5s Show log ›
O
test (sgpu) — success Duration: 1m 30s Show log ›
O
test (mgpu) — success Duration: 45s Show log ›
[dynamo] Support torch.compile with async_compile backend (#188901) X
build — failure Duration: 42s subscription-manager: Remote server error Show log ›
X X X X
[MPS] Add support for torch.linalg.eigh on MPS backend (#188776) O O O O O
Revert: [FSDP2] Enable mixed precision for FSDP2 with autocast (#188650) O O O O O
[export] Handle dynamic shapes in custom ops with structured kernels (#188512) O X
test (cpu) — failure Duration: 2h 1m FAIL: test_overrides (CUDA driver mismatch) Show log ›
O O O
[c10d] Fix ProcessGroupNCCL timeout handling for async ops (#188401) O O O O O
Page 1: Prev | Next

Mockup Pages: Main HUD · CRCR Summary · Per-Repo (PR) · Nightly · Periodic · PR Workflows