Rethink how we identify loops. #101
Merged
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
This commit, small though it is, makes two changes to how we inform
yk about Lua loops:
1. We now calculate where loops start when creating bytecode (rather
than continually examining every opcode in the main loop).
2. We now identify the beginning of the first instruction in a loop as
the start of a loop. Previously we identified the beginning of the
last instruction in a loop as the start of the loop.
The last change can make ~10% difference on some of the benchmarks I
tried, because the trace we get back is both less likely to deopt at the
"wrong points" and is more amenable to trace optimisation.
Contributor
|
LGTM. Will wait for the dependency PR to merge. |
Contributor
Author
|
@vext01 I think we can try this now. |
github-merge-queue
bot
removed this pull request from the merge queue due to no response for status checks
Contributor
Author
|
We might need to wait for the patching patch to be patched in before this patch will merge. |
Contributor
Author
|
Oops, I forgot there's a second version. Forced pushed identical changes to the first. Should hopefully be ready to merge. |
Contributor
|
Are you sure you force pushed? I don't see it. |
Contributor
Author
|
Oops, wrong PR! |
github-merge-queue
bot
removed this pull request from the merge queue due to no response for status checks
Contributor
Author
|
We need the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.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.
[Needs https://github.com/ykjit/yk/pull/1606 merged first.]
This commit, small though it is, makes two changes to how we inform yk about Lua loops:
We now calculate where loops start when creating bytecode (rather than continually examining every opcode in the main loop).
We now identify the beginning of the first instruction in a loop as the start of a loop. Previously we identified the beginning of the last instruction in a loop as the start of the loop.
The last change can make ~10% difference on some of the benchmarks I tried, because the trace we get back is both less likely to deopt at the "wrong points" and is more amenable to trace optimisation.