Store PCall alternate return addresses in frame info #188
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
As suggested by @JohnReppy at POPL'24, one can exploit the static relationship among the return addresses at a PCall to avoid storing the alternate return addresses on the (dynamic) call stack; rather, the `cont`/`seq` return address on the stack should be sufficient to determine the corresponding `parl`/`sync` and `parr`/`spwn` return addresses. In MLton, we can use the `GC_frameInfo` that is associated with each return address to store the `parl`/`sync` and `parr`/`spwn` return addresses in a `PCALL_CONT_FRAME`. One advantage is that it avoids two writes (of the `parl`/`sync` and `parr`/`spwn` return addresses) at each `PCall`. Although it is expected that this overhead is small (the top of the call stack ought to be in cache), it does nicely reduce the overhead of `PCall` to exactly the same as `Call`.
|
Nice! I was able to measure a few percent improvement in a couple examples (fully parallel |
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Outdated suggestions cannot be applied.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.
As suggested by @JohnReppy, rather than storing the alternate return addresses of a
PCallon the (dynamic) call stack, one can store them in a (static) table. In MLton, we can store the alternate return addresses in thestruct GC_frameInfothat corresponds to thePCALL_CONT_FRAME.