Revert "Revert "Qualcomm AI Engine Direct - heap profiling at runtime…#19710
Conversation
… with HTP backend" (pytorch#19705)" This reverts commit 576ed77 which reverted the original heap profiling feature (pytorch#19224). The internal CI test failures that caused the original revert have been fixed by adding @unittest.skip overrides for test_qnn_backend_runtime_option_heap_profile in the simulator test classes. Co-authored-by: Claude <noreply@anthropic.com> Original commit changeset: 576ed77
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19710
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ✅ No FailuresAs of commit b110022 with merge base 2b7a5a2 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
psiddh
requested review from
winskuo-quic
and removed request for
abhinaykukkadapu and
Copilot
meta-cla
Bot
added
the
CLA Signed
psiddh
requested review from
metascroy,
shewu-quic and
winskuo-quic
and removed request for
winskuo-quic
This PR needs a
|
|
@psiddh has imported this pull request. If you are a Meta employee, you can view this in D105884788. |
| ) | ||
| TestQNN.profile_level = 0 | ||
|
|
||
| def test_qnn_backend_runtime_option_heap_profile(self): |
There was a problem hiding this comment.
Hi @psiddh , sorry for the inconvenience caused by the revert PR.
This test is not expected to run on the x86 emulator. Could you help add the following guard? Alternatively, we can also ask @jethroqti to submit the another PR with this fix.
if self.enable_x86_64:
self.skipTest("heap profiling is not supported on host machine")
There was a problem hiding this comment.
Thanks! I already landed an internal fix → adds @unittest.skip overrides in our simulator subclasses, matching the existing .patterns. That said, you can always add this condition as new PR, if you want to. This way any new tests getting added will share the pattern.
There was a problem hiding this comment.
Thanks! We will create another PR to do it.
| ) | ||
| TestQNN.profile_level = 0 | ||
|
|
||
| def test_qnn_backend_runtime_option_heap_profile(self): |
|
Merging this PR now, hopefully the internal CI should be ok this time around |
3 participants
… with HTP backend" (#19705)"
This reverts commit 576ed77 which reverted the original heap profiling feature (#19224). The internal CI test failures that caused the original revert have been fixed by adding @unittest.skip overrides for test_qnn_backend_runtime_option_heap_profile in the simulator test classes.
Original commit changeset: 576ed77