https://stackoverflow.com/questions/2599238/are-memory-barriers-necessary-for-atomic-reference-counting-shared-immutable-dat this answer say that: On x86, it will turn into a lock prefixed assembly instruction, like LOCK XADD. Being a single instruction, it is non-interruptible. As an added "feature", the lock prefix results in a full memory barrier.
In my opinion, full memory barrier means flush store buffer and invalid queue(not very sure, is this right?) On Saturday, September 18, 2021 at 3:17:26 AM UTC+8 Ian Lance Taylor wrote: > On Fri, Sep 17, 2021 at 6:25 AM xie cui <cuiw...@gmail.com> wrote: > > > > how atomic insturction work in golang at x86/amd64, > > I think the atomic insturtion will flush the store buffer and invalid > queue of current cpu core, > > but I am not sure, does someone know about it? > > I assume that you are asking about the sync/atomic package. The > functions in that package will ensure sequential consistency of atomic > operationns, but they will not flush the store buffer. See > https://research.swtch.com/gomm . > > 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/aa7487d1-13f1-451b-86a3-f7f70a7c040cn%40googlegroups.com.
