Thomas was CCed in my original mail actually.:) If I take it correct, with pktgen generating traffic of different flows into internal port, upcalls are expected to be triggered. Is that correct?
Can you help provide any information about the profiling tool used here to obtain the CPU usage of the thread/function? Thanks, Yuxin -----邮件原件----- 发件人: Ben Pfaff [mailto:b...@nicira.com] 发送时间: 2013年12月4日 14:28 收件人: Zhuangyuxin 抄送: discuss@openvswitch.org 主题: Re: ????: [ovs-discuss] Method to verify openvswitch unicast netlink I/O performance I'm surprised you're not CCing Thomas then. Thomas's setup and results look pretty clear to me. pktgen to generate traffic. Without the patch, 74 kpps, with the patch 206 kpps, roughly 3x performance. On Wed, Dec 04, 2013 at 06:22:17AM +0000, Zhuangyuxin wrote: > Actually, the result I mentioned is from Thomas' review. Here's the > whole snippet. I'm wondering how to test and measure the unicast > netlink I/O performance improvement. > > ----------- Cited ----------- > Benchmark > * pktgen -> ovs internal port > * 5M pkts, 5M flows > * 4 threads, 8 cores > > Before: > Result: OK: 67418743(c67108212+d310530) usec, 5000000 (9000byte,0frags) > 74163pps 5339Mb/sec (5339736000bps) errors: 0 > + 2.98% ovs-vswitchd [k] copy_user_generic_string > + 2.49% ovs-vswitchd [k] memcpy > + 1.84% kpktgend_2 [k] memcpy > + 1.81% kpktgend_1 [k] memcpy > + 1.81% kpktgend_3 [k] memcpy > + 1.78% kpktgend_0 [k] memcpy > > After: > Result: OK: 24229690(c24127165+d102524) usec, 5000000 (9000byte,0frags) > 206358pps 14857Mb/sec (14857776000bps) errors: 0 > + 2.80% ovs-vswitchd [k] memcpy > + 1.31% kpktgend_2 [k] memcpy > + 1.23% kpktgend_0 [k] memcpy > + 1.09% kpktgend_1 [k] memcpy > + 1.04% kpktgend_3 [k] memcpy > + 0.96% ovs-vswitchd [k] copy_user_generic_string > > ----------------------------- > > > Thanks, > Yuxin > > -----????????----- > ??????: Ben Pfaff [mailto:b...@nicira.com] > ????????: 2013??12??4?? 13:28 > ??????: Zhuangyuxin > ????: discuss@openvswitch.org > ????: Re: [ovs-discuss] Method to verify openvswitch unicast netlink > I/O performance > > On Wed, Dec 04, 2013 at 02:41:36AM +0000, Zhuangyuxin wrote: > > I noticed there're some patch reviews these days on optimizing > > unicast netlink I/O performance and showed the performance result as > > blow. I'm wondering about the tools and method which we usually use > > to verify performance improvement on netlink IO. Can you help shed > > any light on this? > > I see you provided "before" results, but I don't see any "after" ones. > What were the "after" results? _______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
