-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
O1 fixes. #1400
Merged
Merged
O1 fixes. #1400
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
Merged
Raw numbers below. (The columns don't line up. I know. That's been fixed in a newer rebench here] Before
After
|
ltratt
reviewed
Sep 20, 2024
@vext01 Please squash and rebase against the ykllvm master commit. |
llvm submodule synced/rebased in. |
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
Sep 23, 2024
Please force push an update referencing the new ykllvm commit. |
This change is a collection of fixes required for us to enable the -O1 AOT post-link pipeline in Yk. This is slightly different from just passing -O1, but it's a big step in the right direction. Probably the most significant change is the transition from llvm.experimental.stackamp to llvm.experimental.patchpoint for the control point. This is needed because we were tracking live values at a slightly incorrect point in time. We needed the register state immediately before the call to _ykrt_control_point, but *including* any register shuffling/spills required due to the call ABI. The lua_array_len.c test was removed because we now run enough optimisations for LLVM to statically detect UB in it. Hardly surprising since this was derived using cvise. We can't handle poison values in the JIT and even if we could, I doubt the test is a good one any more. Co-authored-by: Lukas Diekmann <[email protected]>
done |
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.
Requires a compiler change that will come soon.
(Pre-apologies that this isn't all nicely bundled up into self-contained commits -- it's been quite a journey and picking it apart now doesn't seem worthwhile)
This change is a collection of fixes required for us to enable the -O1 AOT post-link pipeline in Yk. This is slightly different from just passing -O1, but it's a big step in the right direction.
Probably the most significant change is the transition from llvm.experimental.stackamp to llvm.experimental.patchpoint for the control point. This is needed because we were tracking live values at a slightly incorrect point in time. We needed the register state immediately before the call to _ykrt_control_point, but including any register shuffling/spills required due to the call ABI.
The lua_array_len.c test was removed because we now run enough optimisations for LLVM to statically detect UB in it. Hardly surprising since this was derived using cvise. We can't handle poison values in the JIT and even if we could, I doubt the test is a good one any more.
Gives between about 12% and 18% speedup on the 4 benchmarks I tried.