iperf3, according to the first sentence on it's home page, is a tool
to measure the bandwidth of a _network_ , not a network node (host).
https://iperf.fr/
Since iperf utilized L4 protocols, which require significant
processing in software, the result depends on the
core and platform speed as well as on the memory bandwidth. If you
run several instantiations of such tests, it is normal that the observed
performance degrades, although both eTSECs operate fully independently.
Some steps that may improve the result include:
1. Affine eTSEC interrupts to CPUs
2. Affine the test tasks to CPUs
3. Disable PTP support
4. Play with interrupt coalescing and Rx/Tx buffer ring sizes.
These parameters are controlled by ethtool
We do not use iperf for benchmarking. There is a simiar test (with similar
limitations), netperf, for which we have a brief reproducibility guide:
https://freescale.sdlproducts.com/LiveContent/content/en-US/QorIQ_SDK/GUID-78005E19-F68E-4169-9828-9...
As of the performance you can expect, benchmarking information requires
NDA and cannot be posted here. Please contact your distributor for
practically obtained performance figures.
Have a great day,
Platon
-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------