Use the custom implementation of multipliedFullWidth on arm64_32 #37905
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
Previously we were falling back on the generic implementation for 64b integers, which resulted in the following codegen: 00000008 asr x8, x0, swiftlang#32 0000000c asr x9, x0, swiftlang#63 00000010 cmp x0, #0x0 00000014 cinv w10, w0, lt 00000018 eor w9, w10, w9 0000001c asr x10, x1, swiftlang#32 00000020 asr x11, x1, swiftlang#63 00000024 cmp x1, #0x0 00000028 cinv w12, w1, lt 0000002c eor w11, w12, w11 00000030 umull x12, w11, w9 00000034 mul x11, x11, x8 00000038 add x11, x11, x12, lsr swiftlang#32 0000003c asr x12, x11, swiftlang#63 00000040 cmp x11, #0x0 00000044 cinv w13, w11, lt 00000048 eor w12, w13, w12 0000004c madd x9, x9, x10, x12 00000050 mul x8, x10, x8 00000054 add x8, x8, x11, asr swiftlang#32 00000058 add x0, x8, x9, asr swiftlang#32 0000005c ret Instead, we should use the 64b implementation when targeting arm64_32, which allows us to generate: 00000008 smulh x0, x1, x0 0000000c ret Unsurprisingly, this is considerably faster.
|
@swift-ci please test |
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Outdated suggestions cannot be applied.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.
Previously we were falling back on the generic implementation for 64b integers, which resulted in the following codegen:
Instead, we should use the 64b implementation when targeting arm64_32, which allows us to generate:
Unsurprisingly, this is considerably faster, though I don't think it will show up in any existing benchmarks.