[AutoScheduler] Use a smaller iteration number for GA to acclerate the search #6994
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
comaniac
reviewed
Nov 29, 2020
There was a problem hiding this comment.
I can give it a try, but it should be fine as long as it's configurable. One related issue is that it's not trivial to configure policy parameters in task scheduler. If we can expose them to task scheduler, I think it's fine to change the default value of most parameters to smaller numbers.
|
@comaniac I added a new argument to control the parameters of search policy from the task scheduler. I tested |
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Dec 2, 2020
…e search (apache#6994) * [AutoScheduler] Use a smaller GA iteration number * fix * fix * add a new argument to control the search policy from task scheduler
trevor-m
pushed a commit
to trevor-m/tvm
that referenced
this pull request
Dec 4, 2020
…e search (apache#6994) * [AutoScheduler] Use a smaller GA iteration number * fix * fix * add a new argument to control the search policy from task scheduler
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Dec 4, 2020
…e search (apache#6994) * [AutoScheduler] Use a smaller GA iteration number * fix * fix * add a new argument to control the search policy from task scheduler
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Outdated suggestions cannot be applied.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.
The default value of
evolutionary_search_num_itersis 10. I did some experiments and found this value is too large.I changed it to 3 or 4 and got almost the same final latency for resnet-50. So I think we can change it to a smaller number to save some search time.
cc @comaniac @jcf94 Could you also verify this for your workloads?