On 05/15/2012 11:47 PM, Voznesensky Vladimir wrote: > Hello. > > I'm doing corosync 2.0.0 cpgbench and cpgbenchzc tests on a single node. > What do I see: > cpgbench > 624586 messages received 8000 bytes per write 10.000 Seconds runtime > 62458.638 TP/s 499.669 MB/s. > cpgbenchzc > 434134 messages received 8000 bytes per write 10.000 Seconds runtime > 43413.357 TP/s 347.307 MB/s. > > Zero-copy code gives worse performance than non-zc one. How could it be? > We had planned to deprecate the zero copy api since the regular API is about 15x more performant in 2.0. Regards -steve > Thanks. > > - VV > _______________________________________________ > discuss mailing list > discuss@xxxxxxxxxxxx > http://lists.corosync.org/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss@xxxxxxxxxxxx http://lists.corosync.org/mailman/listinfo/discuss
