Skip to content

JIT: speed up VNPairForFunc #72527

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.

Sign up for GitHub

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

Jump to bottom
Merged
merged 1 commit into from
Jul 21, 2022
Merged

JIT: speed up VNPairForFunc #72527

merged 1 commit into from
Jul 21, 2022

Conversation

AndyAyersMS
Copy link
Member

No need to call VNForFunc twice, if liberal and conservative VNs agree.

@AndyAyersMS
JIT: speed up VNPairForFunc
8069db4 
No need to call VNForFunc twice, if liberal and conservative VNs agree.
@ghost ghost assigned AndyAyersMS Jul 20, 2022
@ghost ghost added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jul 20, 2022
@ghost
Copy link

ghost commented Jul 20, 2022

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Issue Details

No need to call VNForFunc twice, if liberal and conservative VNs agree.

Author: AndyAyersMS
Assignees: AndyAyersMS
Labels:

area-CodeGen-coreclr
Milestone: -

@AndyAyersMS
Copy link
Member Author

No diffs, locally shows ~ -0.15% TP win.

@jakobbotsch PTAL
cc @dotnet/jit-contrib

@AndyAyersMS AndyAyersMS mentioned this pull request Jul 20, 2022
Merged
@AndyAyersMS
Copy link
Member Author

Seems like there is more that could be done here.

For instance, I don't see why VNForFunc tries optimizations before it consults the cache, and then doesn't cache the optimized results; seems like cache lookup would be way faster than all the case analysis needed for the optimizations.

We should also confirm we're getting decent hash distributions, the hash functions look alarmingly simple.

@AndyAyersMS
Copy link
Member Author

AndyAyersMS commented Jul 20, 2022 •
edited
Loading

X86/ARM SPMI replay legs timed out. Libraries failure is likely unrelated. X86/ARM had no asm diffs and saw TP increase. So it's odd that replay timed out.

Am going to retry these I guess.

@AndyAyersMS
Copy link
Member Author

AndyAyersMS commented Jul 20, 2022 •
edited
Loading

x86 replay succeeded but hit the error where the artifacts exist. Going to ignore this.

@AndyAyersMS AndyAyersMS merged commit 0d96247 into dotnet:main Jul 21, 2022
@AndyAyersMS AndyAyersMS mentioned this pull request Jul 21, 2022
Merged
@ghost ghost locked as resolved and limited conversation to collaborators Aug 20, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

@jakobbotsch jakobbotsch jakobbotsch approved these changes

Assignees

@AndyAyersMS AndyAyersMS

Labels

area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@AndyAyersMS @jakobbotsch