test(rocsparse): mark csrilu0/csric0/bsric0_graph_test as known_bug#9271
Merged
Conversation
jsandham
approved these changes
Jul 10, 2026
✅ All Checks Passed — Ready for Review
📖 Need help? See the Policy FAQ for details on every check and how to fix failures. |
|
🎉 All checks passed! This PR is ready for review. |
Codecov Report✅ All modified and coverable lines are covered by tests. ❌ Your project status has failed because the head coverage (76.92%) is below the target coverage (80.00%). You can increase the head coverage or adjust the target coverage. Additional details and impacted files@@ Coverage Diff @@
## develop #9271 +/- ##
========================================
Coverage 64.99% 64.99%
========================================
Files 2685 2685
Lines 422111 422111
Branches 62707 62707
========================================
Hits 274339 274339
Misses 127073 127073
Partials 20699 20699
*This pull request uses carry forward flags. Click here to find out more. 🚀 New features to boost your workflow:
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Move the hipGraph solve/factor
graph_testcases (thegraph_test: truevariants) of
bsric0,csric0, andcsrilu0tocategory: known_bug. Onmemory-constrained GPUs such as the gfx110X family
(gfx1100/gfx1101/gfx1102/gfx1103), these cases fail with
hipErrorOutOfMemoryin
hipGraphLaunch, which repeatedly blocksrocm-libraries→ROCm/TheRockbump PRs.
Why
capture/launch path OOMs, because it keeps a second copy of the test's device
working set resident across capture + launch, roughly doubling peak VRAM on
these small cards.
allocate a singularity/pivot position buffer via
rocsparse_hipMallocAsync(through
position_t::create_position_async) at solve/factorization time.During hipGraph recording that allocation is captured into the graph and freed
on
hipGraphExecDestroy, causing a use-after-free / OOM on the next graphlaunch.
ROCm/TheRock(e.g. https://github.com/ROCm/TheRock/actions/runs/29073564783).
known_bugin [rocsparse] mark hipGraph solve/factor tests as known_bug #8111, but that was revertedin [rocsparse] Re-enable hipGraph solve/factor tests (revert #8111) #8430 ("Re-enable hipGraph solve/factor tests"), so they run in
pre_checkinagain and OOM.bsrilu0_graph_testwas already re-disabled intest(rocsparse): mark bsrilu0_graph_test as known_bug #9191; this PR extends the same classification to
bsric0,csric0, andcsrilu0.Temporary workaround
This does not fix the underlying memory usage of the graph capture/launch
path — it only unblocks CI. Restoring each test is a one-line change back to
category: pre_checkin. #9143 tracks the real fix.Trade-off / alternative considered
known_bugdisables thesegraph_testcases on all architectures and OSes,including the ones where they currently pass, so that coverage is lost until the
tests are re-enabled. The alternative is a gfx110X-only exclusion via the
skip_hardwareYAML option, which preserves coverage everywhere except theaffected cards; it was not used here because a partial-prefix
skip_hardwarevalue is non-standard for this repo. Happy to switch to that approach instead if
maintainers prefer.
Test plan
test_bsric0.yaml,test_csric0.yaml, andtest_csrilu0.yaml; no source code changed (no new unit test required per repopolicy).
known_bugcases are excluded from the standard/pre_checkin suites, so thegfx110X OOM no longer fails the run.
Made with Cursor