Skip to content

Commutative morph optimizations #64122

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
BruceForstall merged 4 commits into from
Jan 28, 2022
Merged

Commutative morph optimizations #64122

BruceForstall merged 4 commits into from
Jan 28, 2022

Conversation

@SingleAccretion
Copy link
Contributor

@SingleAccretion SingleAccretion commented Jan 21, 2022 •
edited
Loading

As promised in #63251, there was some low-hanging fruit in the code. This change picks it:

  1. Make better use of optimization order.
  2. Fold x + 0 for CNS_LNG too.
  3. Drop one unnecessary fgGlobalMorph guard.
  4. Finally, disable the "mulshift" optimization on non-XARCH targets.

Diffs. Win-x64 diffs didn't make it for unclear reasons, here's the locally obtained version: win-x64 diffs.

@ghost ghost added area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI community-contribution Indicates that the PR has been added by a community member labels Jan 21, 2022
@ghost
Copy link

ghost commented Jan 21, 2022

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

Issue Details

As promised in #63251, there was some low-handing fruit in the code. This change picks it up:

  1. Make better use of optimization order.
  2. Fold x + 0 for CNS_LNG too.
  3. Drop one unnecessary fgGlobalMorph guard.
  4. Finally, disable the "mulshift" optimization on non-XARCH targets.

We are expecting some pretty nice diffs form all the above combined.

Author: SingleAccretion
Assignees: -
Labels:

area-CodeGen-coreclr
Milestone: -

@SingleAccretion SingleAccretion changed the title Commutative morph opt Commutative morph optimizations Jan 21, 2022
@SingleAccretion SingleAccretion marked this pull request as ready for review January 22, 2022 12:48
@SingleAccretion
Copy link
Contributor Author

SingleAccretion commented Jan 22, 2022

@dotnet/jit-contrib

@JulieLeeMSFT JulieLeeMSFT added this to the 7.0.0 milestone Jan 25, 2022
@SingleAccretion
Optimize order of morphing
a4a221c 
"fgMorphCommutative" exposes information to the later
transforms that make them more precise.

For example, "MUL(MUL(X, CONST1), CONST2)" is now morphed
properly into a single "mul" instead of one "mul" plus a shift
and lea in case "CONST2" was a viable candidate for "mulshift".

Another example is how "fgMorphCommutative" can end up with an
"ADD(X, 0)" that was only being discarded in lowering.
@SingleAccretion
Fold (x + 0) => 0 for LONGs on 32 bit
c35e310
@SingleAccretion
Enable one opt outside global morph
6a8543e 
No reason not to. Just a few diffs.
@SingleAccretion
Disable mulshift on ARM/64
d3b168b 
No LEAs - no point.
@BruceForstall BruceForstall merged commit d06d9c9 into dotnet:main Jan 28, 2022
@SingleAccretion SingleAccretion deleted the Commutative-Morph-Opt branch January 28, 2022 20:24
@kunalspathak kunalspathak mentioned this pull request Feb 24, 2022
Closed
@ghost ghost locked as resolved and limited conversation to collaborators Feb 28, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

1 more reviewer

@BruceForstall BruceForstall BruceForstall approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

@BruceForstall BruceForstall

@SingleAccretion SingleAccretion

Labels

area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI community-contribution Indicates that the PR has been added by a community member

Projects

None yet

Milestone

7.0.0

Development

Successfully merging this pull request may close these issues.

3 participants

@SingleAccretion @BruceForstall @JulieLeeMSFT