Cap hexagon max con to number of HVX units #12394
Merged
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
…VX units available.
kparzysz-quic
approved these changes
Aug 12, 2022
csullivan
approved these changes
Aug 12, 2022
xinetzone
pushed a commit
to daobook/tvm
that referenced
this pull request
Nov 25, 2022
…VX units available. (apache#12394)
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Outdated suggestions cannot be applied.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.
Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from Reviewers by @ them in the pull request thread.
Background
Creating more threads than there are HVX units causes performance degradation when threads are unable to reserve an HVX unit and end up running vector instructions on the Scalar cores. Capping the max number of threads to be the number of available HVX units on the device prevents this from occurring and results in better performance on hexagon.
Testing
Ran tests in simulator mode as well on a hexagon HDK to ensure that the expected result of locking max_concurrency to 4 was achieved.