Parallelize optimized op_log_softmax#12099
Conversation
Straightforward application of parallel_for. Differential Revision: [D76831122](https://our.internmc.facebook.com/intern/diff/D76831122/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/12099
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 2 Cancelled JobsAs of commit bd32947 with merge base bed504e ( NEW FAILURE - The following job has failed:
CANCELLED JOBS - The following jobs were cancelled. Please retry:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Straightforward application of parallel_for. Differential Revision: [D76831122](https://our.internmc.facebook.com/intern/diff/D76831122/) ghstack-source-id: 293375210 Pull Request resolved: #12099
|
This pull request was exported from Phabricator. Differential Revision: D76831122 |
|
pull / android / run-emulator is consistently telling us this is broken, but the bug is not obvious to me on rereading |
|
after rereading the code twice, I was inclined to suspect that the kernel is somehow thread-unsafe, so I built with TSAN locally (-fsanitize=debug, CMAKE_BUILD_TYPE=RelWithDebInfo so parallel_for will actually parallelize) and didn't catch a failure. doesn't look like a pre-existing issue on HUD or anything so we shouldn't bypass, but I'm not sure what to do next. Nobody is particularly asking for parallelization AFAIK so there's no rush... |
Straightforward application of parallel_for. Differential Revision: [D76831122](https://our.internmc.facebook.com/intern/diff/D76831122/) [ghstack-poisoned]
Pull Request resolved: #12099 Straightforward application of parallel_for. ghstack-source-id: 293901059 @exported-using-ghexport Differential Revision: [D76831122](https://our.internmc.facebook.com/intern/diff/D76831122/)
|
This pull request was exported from Phabricator. Differential Revision: D76831122 |
|
android / run-emulator is no longer complaining. I guess that means this is landable! |
|
mobilebert has been timing out on HUD, bypassing |
f4f220d
into
gh/swolchok/486/base
This landed internally and PR #12099 closed, but the bot couldn't pick it. Here's a manual pick. Differential Revision: D76831122
This landed internally and PR pytorch#12099 closed, but the bot couldn't pick it. Here's a manual pick. Differential Revision: D76831122
Stack from ghstack (oldest at bottom):
Straightforward application of parallel_for.
Differential Revision: D76831122