12/28/2022 0 Comments Netmap trace route![]() ![]() + Netgate-igb, 351: inet4 packets-per-second X Netgate-igb, 511 (default): inet4 packets-per-second Netmap trace route Pc#On a PC Engines APU2, there is +3% performance benefit + PC-Engines-APU2-igb, 351: inet4 packets-per-second X PC-Engines-APU2-igb, 511 (default): inet4 packets-per-second _bin: 00101011111Īnd we can notice on forwarding performance of a FreeBSD 11.1: _symbolic: ,SWI,NET_NG,NET_TUN,MOUSE,KEYBOARD,ATTACH,CACHED _bin: 00111111111īy replacing this mask by 351, we exclude INTERRUPT and NET_ETHER: _symbolic: ,SWI,INTERRUPT,NET_NG,NET_ETHER,NET_TUN,MOUSE,KEYBOARD,ATTACH,CACHED (Student's t, pooled s = 118987) fastforwardingīy default the binary mask 511 select almost all these source as entropy sources: * HT on, 16q: inet4 packets-per-second forwarded + HT off, 8q: inet4 packets-per-second forwarded ![]() X HT on, 8q (default): inet4 packets-per-second forwarded There is a benefit of about 24% to disable hyper threading on this old CPU.īut here is another example where there is a benefit to kept it enabled (and with the NIC configured to uses all the treads) on Xeon E5 2650L (10c, 20t): * HT-disabled-8rxq: inet packets-per-seconds forwarded + HT-enabled-16rxq: inet packets-per-second forwarded X HT-enabled-8rxq(default): inet packets-per-second forwarded Here is an example on a Xeon E5 2650 (8c,16t) and 10G Chelsio NIC where it improve performance by disabling HT: Netmap trace route drivers#īy default a multi-queue NIC drivers create one queue per core.īut on some older CPU (like Xeon E5-2650 V1) those logical cores didn't help at all for managing interrupts generated by high speed NIC.Įcho 'machdep.hyperthreading_allowed="0"' > /boot/nf On this example we can see that all flows are correctly shared between each 8 queues (about 340K paquets-per-seconds for nic-queue-usage cxl0 Then generating only one flow will use only one NIC queue.ĭuring your load, check that each queues are used with sysctl or with python script like this one that will display real-time usage of each queue. Multi-queue NIC uses feature like Toeplitz Hash Algorithm that balance multiples flows between all cores. Don't try to bench a router with only one flow (same source|destination address and same source|destination port): You need to generate multiples flows. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |