On Fri, Jun 17, 2016 at 2:23 PM, Dmitry Orlov <dmitry.or...@mixpanel.com> wrote: > If I understand it correctly, go scheduler instruments code with additional > yield points not related to I/O. That can help fairness too.
It will sometimes pre-empt on function calls, but this can get a bit hairy since you can't always be sure what function calls will be inlined. This has caused problems for me before where I was adding lots of CPU intensive goroutines to the schedulers queue, and it was context switching so much that it never finished any of them (which I suppose shouldn't have suprised me, but it would be an easy mistake for a new user who may expect CPU intensive work not to yield to make). For example, see this (not-very-scientific) analysis of some bcrypt benchmarks (TL;DR they get faster if we make inlining a lot more aggressive) which has many thousands of function calls in a loop in the blowfish block cipher: http://bl.ocks.org/SamWhited/ebe4f5923526c0d9220f1b5b23b56eba —Sam -- Sam Whited pub 4096R/54083AE104EA7AD3 https://blog.samwhited.com -- 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.
