On Wed, Jun 17, 2009 at 4:40 PM, Tharindu Rukshan Bamunuarachchi<tharindu@xxxxxxxxxxx> wrote: > with Oprofile, due to performance hit, we did not see the issue. at > least we could not pump desired transaction rate. > > however, i could find workaround for 100% issue. > > i have installed two 1G network cards and used bondding with > balance-rr (round robin). > Yes for 1G I think it is resolvable, but for 10G u may be hitting a hardware limit...need to do some calculation: http://seminar.oss.org.tw/fhec/download/Linux%20Multiqueue%20Network%20Driver%20Infrastructure_Herbert%20Xu.pdf But for 10G, u may need hardware feature like TCP Offload Engine (Pradeep S Rautela can describe more about that?) THere is a paper "Performance Characterization of a 10-Gigabit Ethernet TOE" ....check google. > now i do not see 100% utilization of ksoftirqd thread. > > BTW, i heard -rt patch divide ksoftirqd into two pieces for rx and tx. > have you guys -rt patch ? > > > -- Regards, Peter Teoh -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
