[Mosaic TPU] Fold reshape (..., M, N, 128) -> (..., M, N * 128) with any tiling and dtype to load. #28309
Conversation
This file contains hidden or 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
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
copybara-service
bot
force-pushed
the
test_751696195
branch
from
472645d to
24d051a
Compare
…any tiling and dtype to load. Now we should support decent number of reshape that match with this pattern. And it is much more efficient to fold reshape (retiling + re-pack) into load. Take bf16(256, 8, 128) -> bf16(256, 8 * 128) as example, this cl emits 148 bundles and compared to before 630 bundles - about 4.2x speedup. PiperOrigin-RevId: 751696195
copybara-service
bot
force-pushed
the
test_751696195
branch
from
24d051a to
fd75522
Compare
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Outdated suggestions cannot be applied.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.
[Mosaic TPU] Fold reshape (..., M, N, 128) -> (..., M, N * 128) with any tiling and dtype to load.
Now we should support decent number of reshape that match with this pattern. And it is much more efficient to fold reshape (retiling + re-pack) into load.
Take bf16(256, 8, 128) -> bf16(256, 8 * 128) as example, this cl emits 148 bundles and compared to before 630 bundles - about 4.2x speedup.