[dpdk-users] testpmd packet drop issue at 4.5 million pps

kamal sivadas kamalsivadas01 at gmail.com
Sat Jul 9 16:55:02 CEST 2016


Hi All,

I am trying to run testpmd on Intel-XL710 NIC with one queue. I am seeing
constant RX drops. Can someone help me to identify the problem here.

System info: Cent OS 7. CPU optimized for performance. I followed
http://dpdk.org/doc/guides-16.04/linux_gsg/nic_perf_intel_platform.html

Traffic flow: 27Gbps, 4.3 million pps(average packet size is 780 bytes)

DPDK version: 2.2



./testpmd -l 4,5 -n 4 -- -i --portmask=0x1 --numa --nb-cores=1 --rxd=2048
--rss-udp


Note: I tried with default RX descriptor size and that doesn’t work, so
increased to 2048.

Increasing the number of forwarding core to 2 with 2 queue also didn’t help.

This is another command line option I tried but didn't help

./testpmd -l 4,5,6 -n 4 --socket-mem 2048,0 -- -i --port-numa-config=0,1
--socket-num=0  --portmask=0x1 --mbcache=512  --nb-cores=2 --rxq=2 --txq=2
--rxd=2048 --rss-udp --forward-mode=io --port-topology=chained --rxpt=32
--rxht=32 --rxfreet=32 --rxwt=32




Test pmd start logs:

io packet forwarding - CRC stripping disabled - packets/burst=32

  nb forwarding cores=1 - nb forwarding ports=1

  RX queues=1 - RX desc=2048 - RX free threshold=32

  RX threshold registers: pthresh=8 hthresh=8 wthresh=0

  TX queues=1 - TX desc=512 - TX free threshold=32

  TX threshold registers: pthresh=32 hthresh=0 wthresh=0

  TX RS bit threshold=32 - TXQ flags=0xf01


Any assist would be really helpful.


Regards,

Kamal


More information about the users mailing list