[AutoScheduler] Fix the conflict of thread pool in measurement #7166
Conversation
This file contains 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
merrymercy
force-pushed
the
pr-fix-measure
branch
2 times, most recently
from
2070e6b to
ad639bb
Compare
merrymercy
force-pushed
the
pr-fix-measure
branch
from
ad639bb to
de601aa
Compare
jcf94
approved these changes
Dec 25, 2020
| """Call a function with timeout""" | ||
| que = multiprocessing.Queue(2) | ||
| process = multiprocessing.Process(target=_func_wrapper, args=(que, func, args, kwargs)) | ||
| process = multiprocessing.Process( |
There was a problem hiding this comment.
I'm thinking about whether the bug was introduced by the multiprocessing? Maybe directly use thread here can solve the problem.... I'm not sure, just a guess.
There was a problem hiding this comment.
We can only use Process here. threading.Thread does not support timeout when calling a c++ function.
|
Thanks @merrymercy @jcf94 |
tkonolige
pushed a commit
to tkonolige/incubator-tvm
that referenced
this pull request
Jan 11, 2021
TusharKanekiDey
pushed a commit
to TusharKanekiDey/tvm
that referenced
this pull request
Jan 20, 2021
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Jan 21, 2021
electriclilies
pushed a commit
to electriclilies/tvm
that referenced
this pull request
Feb 18, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Currently, due to the conflict between python' multiprocessing and tvm's thread pool, we cannot run auto-scheduler search after running any tvm-generated functions.
I found this conflict can be avoided if we run the measurement function in
LocalRunnerwithin a newthreading.Thread.This PR just added such a new wrapper in
auto_scheduler.local_runner.run. With this PR, we can freely run auto-scheduler search after executing any tvm-generated functions or relay's FoldConstant.