On Wed, 27 Jul 2016 08:02:51 -0700 (PDT) Alexey Solovey <acte...@gmail.com> wrote:
> Hello. Will support of WebAssembly (compilation to JS) in Go > language, in ending version? If by "ending version" you mean 1.7 which is now at RC3, then AFAIK the answer is no -- see for youself: https://tip.golang.org/doc/go1.7 > Hardness: SIMD, Atomics, Threads... I can't quite parse this. Do you mean these things are hard to implement for WebAssembly? If yes, then I doubt so; here's why. SIMD is an implementation detail as nothing in the Go language spec talks about SIMD. Explicit support for atomic access to values is really platform specific: there may exist platforms on which the stuff provided by the sync/atomic package will all be no-ops, and WebAssembly could well be such a platform. Go does not support threads. I mean, its runtime actually requires support for OS threads to do useful things, but threads are not exposed to the language (there's nothing in the Go language spec which talks about threads). > Also, I have question: GO supports SIMD? If you mean auto-vectorization of certain calculations by the compiler then IIUC the "stock" Go distribution (that one, originated from Google, dubbed "gc" in the community) does not have it (yet, at least) but the "gccgo" implementation which is a front-end to GCC piggybacks on the GCC's low-level optimizer and the machine code generation backends which _could_ do this. There's also an ongoing project to implement Go front-end for LLVM, and that one might have such support as well. -- 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. For more options, visit https://groups.google.com/d/optout.
