[Hexagon] Add scripts for e2e MetaSchedule tuning demonstration #13135
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
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment. Generated by tvm-bot |
areusch
added
needs-triage
csullivan
approved these changes
Oct 20, 2022
Comment on lines
+113
to
+117
| print( | ||
| "max and mean abs difference with the reference:", | ||
| np.max(np.abs(ref_result - hexagon_output)), | ||
| np.mean(np.abs(ref_result - hexagon_output)), | ||
| ) |
There was a problem hiding this comment.
Are there accuracy issues with tuning the fp16 variant that prevent an assert_allclose with the reference?
There was a problem hiding this comment.
There is non-trivial accuracy difference between x86 and HVX fp16. Added a sample output and assert check with a fairly loose bound.
xinetzone
pushed a commit
to daobook/tvm
that referenced
this pull request
Nov 10, 2022
…he#13135) I've worked on a series of PRs to enable e2e MS tuning on Hexagon (mostly for supporting link-params = True in MS). Now that all the pieces have been upstreamed, I'm adding demo tuning scripts under test_hexagon/metaschedule_e2e. They are not run on CI, and running it locally requires PyTorch to generate fp16 and int8 resnet50. The scripts use a small number of tuning trials and replay-trace search strategy instead of the evolutionary search, to finish tuning quickly. Those interested in MS tuning can tweak these settings for better performance at the cost of more tuning times. * change dtype in tvmscript roundtrip test to avoid int printing error * allow printing non int8 array * Revert "change dtype in tvmscript roundtrip test to avoid int printing error" * add loose assert check on fp16 result
xinetzone
pushed a commit
to daobook/tvm
that referenced
this pull request
Nov 25, 2022
…he#13135) I've worked on a series of PRs to enable e2e MS tuning on Hexagon (mostly for supporting link-params = True in MS). Now that all the pieces have been upstreamed, I'm adding demo tuning scripts under test_hexagon/metaschedule_e2e. They are not run on CI, and running it locally requires PyTorch to generate fp16 and int8 resnet50. The scripts use a small number of tuning trials and replay-trace search strategy instead of the evolutionary search, to finish tuning quickly. Those interested in MS tuning can tweak these settings for better performance at the cost of more tuning times. * change dtype in tvmscript roundtrip test to avoid int printing error * allow printing non int8 array * Revert "change dtype in tvmscript roundtrip test to avoid int printing error" * add loose assert check on fp16 result
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.
I've worked on a series of PRs to enable e2e MS tuning on Hexagon (mostly for supporting
link-params = Truein MS). Now that all the pieces have been upstreamed, I'm adding demo tuning scripts undertest_hexagon/metaschedule_e2e.They are not run on CI, and running it locally requires PyTorch to generate fp16 and int8 resnet50.
The scripts use a small number of tuning trials and
replay-tracesearch strategy instead of the evolutionary search, to finish tuning quickly. Those interested in MS tuning can tweak these settings for better performance at the cost of more tuning times.@csullivan @kparzysz-quic @farshidsp