Defer global caching of CodeInstance to post-optimization step
#58343
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
In certain cases, the optimizer can introduce new type information. This is particularly evident in SROA, where load forwarding can reveal type information that was not visible during abstract interpretation. In such cases, re-running abstract interpretation using this new type information can be highly valuable, however, currently, this only occurs when semi-concrete interpretation happens to be triggered. This commit introduces a new "post-optimization inference" phase at the end of the optimizer pipeline. When the optimizer derives new type information, this phase performs IR abstract interpretation to further optimize the IR.
vtjnash
reviewed
May 7, 2025
vtjnash
reviewed
May 7, 2025
vtjnash
approved these changes
May 7, 2025
Keno
approved these changes
May 8, 2025
aviatesk
approved these changes
May 8, 2025
vtjnash
added
compiler:inference
serenity4
added a commit
to serenity4/Cthulhu.jl
that referenced
this pull request
May 9, 2025
serenity4
added a commit
to serenity4/Cthulhu.jl
that referenced
this pull request
May 9, 2025
serenity4
added a commit
to serenity4/Cthulhu.jl
that referenced
this pull request
May 9, 2025
serenity4
added a commit
to JuliaDebug/Cthulhu.jl
that referenced
this pull request
May 9, 2025
* Adjust to JuliaLang/julia#58343 * Make branch static * Bump version
serenity4
added a commit
to serenity4/Diffractor.jl
that referenced
this pull request
May 9, 2025
Keno
pushed a commit
to JuliaDiff/Diffractor.jl
that referenced
this pull request
May 9, 2025
58 tasks
charleskawczynski
pushed a commit
to charleskawczynski/julia
that referenced
this pull request
May 12, 2025
…iaLang#58343) This PR extracts the caching improvements from JuliaLang#56687, implemented by @aviatesk. It essentially defers global caching to the post-optimization step, giving a temporary cache to the optimizer instead of relying on the global cache. The issue with caching globally before optimization is that any errors occurring within the optimizer may leave a partially initialized `CodeInstance` in the cache, which was meant to be updated post-optimization. Exceptions should not be thrown in the native compilation pipeline, but abstract interpreters extending optimization routines will frequently encounter some during an iterative development (see JuliaComputing/DAECompiler.jl#25). An extra benefit from deferring global caching is that the optimizer can then more safely update `CodeInstance`s with new inference information, as is the intent of JuliaLang#56687. --------- Co-authored-by: Shuhei Kadowaki <[email protected]> Co-authored-by: Cody Tapscott <[email protected]>
charleskawczynski
pushed a commit
to charleskawczynski/julia
that referenced
this pull request
May 12, 2025
…iaLang#58343) This PR extracts the caching improvements from JuliaLang#56687, implemented by @aviatesk. It essentially defers global caching to the post-optimization step, giving a temporary cache to the optimizer instead of relying on the global cache. The issue with caching globally before optimization is that any errors occurring within the optimizer may leave a partially initialized `CodeInstance` in the cache, which was meant to be updated post-optimization. Exceptions should not be thrown in the native compilation pipeline, but abstract interpreters extending optimization routines will frequently encounter some during an iterative development (see JuliaComputing/DAECompiler.jl#25). An extra benefit from deferring global caching is that the optimizer can then more safely update `CodeInstance`s with new inference information, as is the intent of JuliaLang#56687. --------- Co-authored-by: Shuhei Kadowaki <[email protected]> Co-authored-by: Cody Tapscott <[email protected]>
|
Doesn't backport cleanly. |
Is it the PR author who backports it in this case? In any case, I'll do this one. |
|
I don't think the backport is necessarily required. |
|
That's true, it can definitely wait for 1.13. |
|
Removed the backport label. |
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 PR extracts the caching improvements from #56687, implemented by @aviatesk. It essentially defers global caching to the post-optimization step, giving a temporary cache to the optimizer instead of relying on the global cache.
The issue with caching globally before optimization is that any errors occurring within the optimizer may leave a partially initialized
CodeInstancein the cache, which was meant to be updated post-optimization. Exceptions should not be thrown in the native compilation pipeline, but abstract interpreters extending optimization routines will frequently encounter some during an iterative development (see JuliaComputing/DAECompiler.jl#25).An extra benefit from deferring global caching is that the optimizer can then more safely update
CodeInstances with new inference information, as is the intent of #56687.