Remove i64.mul128, add i64.mul_wide_{s,u}
#13
Merged
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
Some recent [benchmarking] had a surprising result I wasn't trying to dig for. Notably as summarized in #11 some more use cases of widening multiplication being more optimal than 128-by-128 bit multiplication have started to arise. Coupled with local benchmarking confirming that both on x64 and aarch64 that widening multiplication has more support in LLVM for more optimal lowerings and was easier to implement in Wasmtime than the 128-by-128 bit multiplication once various optimizations were implemented. In the end `i64.mul128`, which was primarily motivated by "feels cleaner" and "should have the same performance" as widening multiplication, does not appear to have the expected performance/implementation tradeoff. Getting an as-performant `i64.mul128` instruction relative to `i64.mul_wide_{s,u}` has required more work than expected and so the balance of concerns has me now tipping away from `i64.mul128`, despite it being "less clean" compared to the add/sub opcodes proposed in this PR. Closes #11 [benchmarking]: #6 (comment)
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Outdated suggestions cannot be applied.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.
Some recent benchmarking had a surprising result I wasn't trying to dig for. Notably as summarized in #11 some more use cases of widening multiplication being more optimal than 128-by-128 bit multiplication have started to arise. Coupled with local benchmarking confirming that both on x64 and aarch64 that widening multiplication has more support in LLVM for more optimal lowerings and was easier to implement in Wasmtime than the 128-by-128 bit multiplication once various optimizations were implemented.
In the end
i64.mul128, which was primarily motivated by "feels cleaner" and "should have the same performance" as widening multiplication, does not appear to have the expected performance/implementation tradeoff. Getting an as-performanti64.mul128instruction relative toi64.mul_wide_{s,u}has required more work than expected and so the balance of concerns has me now tipping away fromi64.mul128, despite it being "less clean" compared to the add/sub opcodes proposed in this PR.Closes #11