Repository navigation
floats v0.10.0: drop the max/min zero fallback and the moved sum/elementwise assembly - #100
Merged
Merged
Conversation
… four-arch elementwise asm - go-simd/floats v0.9.0 -> v0.10.0. - maxSIMD/minSIMD (amd64, arm64) return floats.Max/Min directly: the signed-zero contract is floats' since floats#32, so the serial recomputation of a zero result is gone. - ppc64le, loong64, riscv64, s390x: sqrtSIMD and addBin/subBin/mulBin/divBin call floats.Sqrt/Add/Sub/Mul/Div (floats#30: the kernels moved there with the same instruction sequences and gate themselves on the CPU feature). The float64 sumVSX/sumLASX/sumRVV/sumVX (dead since v0.37.0) and the float64 sqrt/add/sub/mul/div kernels are removed from the asmgen generators and from sum_<arch>.s (regenerated; amd64 and arm64 output unchanged). - amd64 without AVX2+FMA runs floats' SSE2 kernels (floats#31). Native IBM z17 (ssh linuxone): go test ./internal/kernels passes. Max/Min and dot/gemm kernels of the four architectures stay here. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Bumps go-simd/floats to v0.10.0 and uses it:
maxSIMD/minSIMDreturnfloats.Max/Min(signed zeros fixed in floats#32), so the serial fallback for a zero result is removed.sum_<arch>.s(amd64/arm64 output identical).Verified:
go test ./...on arm64,go vetfor the five other targets, native IBM z17go test ./internal/kernels(cross-compiled, scp). The qemu lanes of CI cover ppc64le, loong64, riscv64. Max/Min, dot and GEMM kernels of the four architectures stay in ndarray.🤖 Generated with Claude Code