On Jan 9, 2014, at 12:56 PM, Ethan Jackson <et...@nicira.com> wrote:
> FWIW the exact percentage gains are different in my setup, but the > general result that these two patches stacked on top of each other > make a big difference is consistent with what I've seen. > I had to bump the number of netperfs up to 50, and I have seen the flow stats reflect a significantly higher number of packets/second as a result. Method: take two stats and compute the diffs in time and number of packets and bytes, divide to get packets/sec. I suspect that the reported flow duration will not exactly match the time the flow was last dumped (does it?) so there likely is rather large inaccuracy on these numbers. But they seem to go hand in hand with the number of TCP connection setups as they should. Jarno > Ethan > > On Thu, Jan 9, 2014 at 12:41 PM, Ben Pfaff <b...@nicira.com> wrote: >> On Thu, Jan 09, 2014 at 11:50:31AM -0800, Jarno Rajahalme wrote: >>> I tested this (without the extra malloc), and this is a minor >>> improvement on master, but a huge, in the order of ~40% performance >>> increase on top of the NUMA patches I sent yesterday. Also, >>> performance does not vary so much any more, and it is now actually >>> beneficial to add more threads and netperfs. Now, with this and the >>> NUMA level kernel stats, find_match_wc() is actually the top line on >>> ?perf?, as it should! >>> >>> Based on testing only: >>> Acked-by: Jarno Rajahalme <jrajaha...@nicira.com> >> >> That is awesome. >> >> I'm still working on fixing all the races in initialization and >> destruction of fatlocks. I'll send it out for final review when I'm >> happy. >> >> Thanks, >> >> Ben. >> _______________________________________________ >> dev mailing list >> dev@openvswitch.org >> http://openvswitch.org/mailman/listinfo/dev _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
