analyse and trend outputs of vpp "show run" for cycles-per-packet (CPP), m-thread compare:
x520 - 1-worker, 2-worker, 4-worker
xl710 - 1-worker, 2-worker, 4-worker
x710 - 1-worker, 2-worker, 4-worker
vic1227 - 1-worker, 2-worker, 4-worker
vic1385 - 1-worker, 2-worker, 4-worker

[ ] TF, PM multi-thread NDR analysis - verify linearity of NDR subject to NIC and PCIe limits
[ ] NN x520 - 1c, 2c, 4c, watch NIC limit of 23Mpps, (aggr)
[ ] NN x710 - 1c, 2c, 4c, watch NIC limit of ??Mpps (aggr)
[ ] NN xl710 - 1c, 2c, 4c, watch NIC limit of 35.8Mpps, 50Mbps (aggr)
[ ] NN analyse and trend outputs of vpp "show run" for cycles-per-packet (CPP), m-thread compare:
[ ] NN x520 - 1-worker, 2-worker, 4-worker
[ ] NN xl710 - 1-worker, 2-worker, 4-worker
[ ] NN x710 - 1-worker, 2-worker, 4-worker
[ ] NN vic1227 - 1-worker, 2-worker, 4-worker
[ ] NN vic1385 - 1-worker, 2-worker, 4-worker

I did s/??/NN/ per my email to csit-core.

Which data do you want to have on y-axis? average vectors/node?

NDR throughput, as it says in the task - verify linearity of NDR subject to NIC and PCIe limits.
and if PDR is >> NDR, do the same for PDR.

Graphing vector size is pointless for NDR throughput

If it’s still not clear, let’s have a live call for me to walk you thru what I mean, as it’s really important that you understand.