pytorch - ✅(Solved) Fix #178639 breaks profiler-dependent tests on `aarch64` [1 pull requests, 1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
pytorch/pytorch#180002Fetched 2026-04-11 06:10:21
View on GitHub
Comments
0
Participants
1
Timeline
31
Reactions
0
Author
Participants
Timeline (top)
mentioned ×14subscribed ×14labeled ×2cross-referenced ×1

Fix Action

Fixed

PR fix notes

PR #180005: [ARM] Upgrade kineto to fix aarch64 platforms missing events #180002

Description (problem / solution / changelog)

Seems to fix #180002, as proposed by @saagarjha

cc @robieta @chaekit @guotuofeng @guyang3532 @dzhulgakov @davidberard98 @briancoutinho @sraikund16 @sanrise @mwootton @divyanshk @jiannanWang @scotts @ryanzhang22 @snadampal @milpuz01 @aditew01 @nikhil-arm @fadara01 @robert-hardwick @nWEIdia

Changed files

  • third_party/kineto (modified, +1/-1)
RAW_BUFFERClick to expand / collapse

🐛 Describe the bug

Observed on tests that rely on the profiler to e.g., get kernel info such as python test/test_foreach.py TestForeachCUDA.test_big_num_tensors__foreach_max_use_cuda_graph_False_w_empty_False_cuda_float32 that results seem to be dropped (e.g., on GB200, thor).

Versions

near top-of-tree source build

cc @robieta @chaekit @guotuofeng @guyang3532 @dzhulgakov @davidberard98 @briancoutinho @sraikund16 @sanrise @mwootton @divyanshk @jiannanWang @scotts @ryanzhang22

extent analysis

TL;DR

Investigate the profiler's behavior in the tests to determine why results are being dropped, particularly for specific hardware configurations like GB200 and thor.

Guidance

  • Review the test cases, such as test_foreach.py, to understand how the profiler is being used and what results are expected.
  • Check the profiler's configuration and settings to see if there are any limitations or constraints that could be causing results to be dropped.
  • Consider running the tests with additional logging or debugging enabled to gather more information about the profiler's behavior.
  • Investigate if the issue is specific to certain hardware configurations, such as GB200 and thor, and if so, determine what differences in these configurations might be contributing to the problem.

Notes

The issue seems to be related to the profiler's behavior in specific test cases and hardware configurations, but without more information, it's difficult to provide a more detailed solution.

Recommendation

Apply workaround: Investigate and adjust the profiler's configuration and settings to ensure that results are not being dropped, as the root cause of the issue is likely related to the profiler's behavior.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

pytorch - ✅(Solved) Fix #178639 breaks profiler-dependent tests on `aarch64` [1 pull requests, 1 participants]