I am greatly in favor of doing an experiment with user-mode scheduling. UMS is this magical thing where the operating system asks you what to do next when it blocks a thread on I/O (or a page fault/yield). I think it would be possible to create a much more efficient thread pool than what we currently have.
So +1 from me; try to implement an UMS-based thread pool and run the 'stat' benchmark so see if it makes for a good improvement. - Bert On Wednesday, June 17, 2015 at 9:43:36 AM UTC-7, Iñaki Baz Castillo wrote: > > 2015-06-16 23:13 GMT+02:00 <[email protected]>: > > Hopefully, with these new options, multicore performance scaling can be > > improved. > > It seems that you are trying to improve something that cannot be > improved given that libuv is mostly single threaded. > > > -- > Iñaki Baz Castillo > <[email protected]> > -- You received this message because you are subscribed to the Google Groups "libuv" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/libuv. For more options, visit https://groups.google.com/d/optout.
