> -----Original Message----- > From: Kevin Traynor [mailto:ktray...@redhat.com] > Sent: Wednesday, April 5, 2017 2:58 PM > To: O Mahony, Billy <billy.o.mah...@intel.com>; Maxime Coquelin > <maxime.coque...@redhat.com>; d...@openvswitch.org > Subject: Re: [ovs-dev] [PATCH] netdev-dpdk: Enable INDIRECT_DESC on > DPDK vHostUser. > > On 03/20/2017 11:18 AM, O Mahony, Billy wrote: > > Hi Maxime, > > > >> -----Original Message----- > >> From: Maxime Coquelin [mailto:maxime.coque...@redhat.com] > >> Sent: Friday, March 17, 2017 9:48 AM > >> To: O Mahony, Billy <billy.o.mah...@intel.com>; d...@openvswitch.org > >> Subject: Re: [ovs-dev] [PATCH] netdev-dpdk: Enable INDIRECT_DESC on > >> DPDK vHostUser. > >> > >> Hi Billy, > >> > >> On 03/01/2017 01:36 PM, Billy O'Mahony wrote: > >>> Hi All, > >>> > >>> I'm creating this patch on the basis of performance results outlined > >>> below. In summary it appears that enabling INDIRECT_DESC on DPDK > >>> vHostUser ports leads to very large increase in performance when > >>> using linux stack applications in the guest with no noticable > >>> performance drop for DPDK based applications in the guest. > >>> > >>> Test#1 (VM-VM iperf3 performance) > >>> VMs use DPDK vhostuser ports > >>> OVS bridge is configured for normal action. > >>> OVS version 603381a (on 2.7.0 branch but before release, > >>> also seen on v2.6.0 and v2.6.1) DPDK v16.11 QEMU v2.5.0 (also > >>> seen with v2.7.1) > >>> > >>> Results: > >>> INDIRECT_DESC enabled 5.30 Gbit/s > >>> INDIRECT_DESC disabled 0.05 Gbit/s > >> This is indeed a big gain. > >> However, isn't there a problem when indirect descriptors are disabled? > >> 0.05 Gbits/s is very low, no? > > > > [[BO'M]] Yes the disabling of the indirect descriptors feature appears to be > what causes the very low test result. And the root cause may actually be > related to this bug > https://bugs.launchpad.net/ubuntu/+source/qemu/+bug/1668829 . > However, turning on the indirect descriptors certainly helps greatly. > > > >> > >> Could you share the iperf3 command line you used? > > > > [[BO'M]] In the server VM "iperf3 -s" and in the client "iperf3 -c <server > > ip > addr> -t 30". -t 30 is the duration (secs) of the test. OVS-DPDK bridge was > set > to use normal action. > > > Hi Billy, > > I ran the iperf test on master and I see very different results than you got? > > mrg on/indirect off: 7.10 Gbps > mrg off/indirect off: 5.05 Gbps > mrg off/indirect on: 7.15 Gbps
[[BO'M]] Hi Kevin, By those figures the performance is still +40% in the right direction for using indirect descriptors. What version of qemu did you use? (if as per the Launchpad bug qemu is a root cause here). Also in that case kernel versions may be significant. I was using qemu 2.5 (tagged release, built locally) and Ubuntu 16.04.01 with 4.04 kernel in the guest. I can retry the tests with head of master when I get a chance but the patch is still offering a large improvement. Cheers, Billy. > > Kevin. > > >> > >>> Test#2 (Phy-VM-Phy RFC2544 Throughput) DPDK PMDs are polling NIC, > >>> DPDK loopback app running in guest. > >>> OVS bridge is configured with port forwarding to VM (via > >>> dpdkvhostuser > >> ports). > >>> OVS version 603381a (on 2.7.0 branch but before release), > >>> other versions not tested. > >>> DPDK v16.11 > >>> QEMU v2.5.0 (also seen with v2.7.1) > >>> > >>> Results: > >>> INDIRECT_DESC enabled 2.75 Mpps @64B pkts (0.176 Gbit/s) > >>> INDIRECT_DESC disabled 2.75 Mpps @64B pkts (0.176 Gbit/s) > >> > >> Is this with 0% packet loss? > > [[BO'M]] Yes. To an accuracy of .05 Mpps. > >> > >> Regards, > >> Maxime > > _______________________________________________ > > dev mailing list > > d...@openvswitch.org > > https://mail.openvswitch.org/mailman/listinfo/ovs-dev > > _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
