struct CaseSet: Optimize by matching on len once per set_ctx
#1283
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 C, `case_set` works by `switch`ing once on `len`, once per `set_ctx`, but my `CaseSet` implementation in Rust accidentally switched to `match`ing on `buf.len()` each time. This goes back to the original, optimal behavior. To do this, `set_ctx` is made a rank-2 polymorphic "closure". This does not exist in Rust, so it's emulated through a generic trait with a generic method. This inner generic over `trait CaseSetter` is what allows `fn CaseSet::one` to select the correct `CaseSetterN` at compile time. However, this means that closures can't be used anymore, which is very annoying. To partially remedy this, I added the `set_ctx!` macro, which emulates the `set_ctx` closure as much as possible. All captures (up vars in `rustc`) and their types must be declared.
kkysen
force-pushed
the
kkysen/case_set-rank2
branch
from
4df8569 to
0de4037
Compare
Collaborator
|
I'm working on validating performance for this. It's not making as much of a difference as expected, and I think some checks are not getting elided. |
Collaborator
Author
😭 I didn't get a chance to look at the asm yet, though. Let me know if there's anything simple we can do there. I can look a bit too to see which things aren't elided. |
Collaborator
|
I have a very different way of doing this that is a lot less complicated, I think, so don't spend time on it for now. |
Collaborator
Author
Oh interesting, hope it works! |
kkysen
changed the title
struct CaseSet: Optimize by matching on len one per set_ctxstruct CaseSet: Optimize by matching on len once per set_ctx
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.
In C,
case_setworks byswitching once onlen, once perset_ctx, but myCaseSetimplementation in Rust accidentally switched tomatching onbuf.len()each time. This goes back to the original, optimal behavior.To do this,
set_ctxis made a rank-2 polymorphic "closure". This does not exist in Rust, so it's emulated through a generic trait with a generic method. This inner generic overtrait CaseSetteris what allowsfn CaseSet::oneto select the correctCaseSetterNat compile time.However, this means that closures can't be used anymore, which is very annoying. To partially remedy this, I added the
set_ctx!macro, which emulates theset_ctxclosure as much as possible. All captures (up vars inrustc) and their types must be declared.I didn't actually benchmark this or look at the asm yet, though, if anyone wants to do that.