FWIW, I took a stab at a SIMD-oriented feature (https://go.dev/issue/48499), but as @Ian%20Lance%20Taylor put it, it's about the right approach. I skewed too far towards convenience in what I proposed, gaining significant maintainability concerns. On Thursday, April 7, 2022 at 3:35:35 PM UTC-5 Ian Lance Taylor wrote:
> On Thu, Apr 7, 2022 at 11:03 AM Arthur Comte <ad...@arthurcomte.com> > wrote: > > > > Intrinsic functions (functions that compile down to one hardware > instruction) are common in (relatively) low-level languages. Go even > supports embed ASM (though a weird flavour of it) > > Can someone explain why we do not have an intrinsic function package? > This seems like it would greatly ease the use of SIMD > > Go does have a couple of intrinsic function packages: sync/atomic and > math/bits. Many of the functions in those packages are implemented > directly in the compiler. > > What Go does not have is a generalized way to write intrinsic > functions apart from implementing them in the compiler. That was > https://go.dev/issue/17373, which was declined. > > Go also does not have a SIMD package, but that is more because nobody > has figured out the right approach for such a thing in Go. There was > at least one attempt at https://go.dev/issue/35307, but that too was > declined. > > Ian > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/92bb2a89-b2ec-46ea-9223-294e963ebbb4n%40googlegroups.com.
