[Optimization] Make Z inline better #6839
Open
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
effectfully
added
Do not merge
EXPERIMENT
effectfully
temporarily deployed
to
github-pages
GitHub Actions
Inactive
effectfully
commented
Feb 12, 2025
| ({cpu: 432732134016 | ||
| | mem: 1722650147}) |
There was a problem hiding this comment.
I'm rather puzzled as to how eta-expanding Z leads to such a massive slowdown.
| 1 |
There was a problem hiding this comment.
The semantics change.
zliu41
reviewed
Feb 12, 2025
effectfully
commented
Feb 13, 2025
effectfully
commented
Feb 13, 2025
Comment on lines
0
to
2
| ({cpu: 5476180 | ||
| | mem: 28020}) No newline at end of file | ||
| ({cpu: 5428180 | ||
| | mem: 27720}) |
Comment on lines
0
to
2
| ({cpu: 6684432 | ||
| | mem: 33922}) No newline at end of file | ||
| ({cpu: 6636432 | ||
| | mem: 33622}) |
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Outdated suggestions cannot be applied.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.
This makes Z inline better, if it's already getting inlined. I.e. if it's already getting inlined, we get a speedup with this PR (and a bit less size consumed), otherwise we get a slowdown (and a bit more size consumed).
If we're gonna start always inlining Z, this is likely beneficial, otherwise not.
The idea is simple: eta-expand Z, so that inlining the stepper function gives us back a value (rather than an application) which can be inlined further. The gist of the change:
This partially cancels what #5939 did and reintroduces the strictness issue that that PR fixed, which is probably not good, but I dunno.
It'd probably be best to just mark
fixas inlinable, since we generate it ourselves anyway and can do it. Then we won't need to rely on it being defined a certain way.