Orin‑Nano PCI‑E Ethernet fails to reach expected throughput in iperf3 TCP bidirectional test

[3285195874]

Module: Jetson Orin‑Nano 4GB

L4T Version: R36.4.4

Link: Ethernet, link negotiated at 2500M Full Duplex with auto‑negotiation enabled.

In TCP bidirectional iperf test (Orin-Nano): Aggregate total throughput only reaches around 1.81Gbps, cannot hit the full 2.5G line-rate capacity.

Are there any recommended tuning methods for PCIe Ethernet on Orin‑Nano R36.4.4 to achieve aggregate TCP bidirectional throughput above 2300 Mbps?

In addition，I tried to modify the interrupt affinity for this PCIe NIC. When executing echo 2 > /proc/irq/xxx/smp_affinity, it returns Invalid argument.

[WayneWWW]

Does the cpu0 loading already get the 100% when you do the test?

[3285195874]

CPU0 does not reach 100% utilization, and overall total CPU usage is also not high.

[3285195874]

I tried to modify the interrupt affinity for this PCIe NIC. When executing echo 2 > /proc/irq/xxx/smp_affinity, it returns Invalid argument.

@3285195874

Do you have any ways to solve this?

[WayneWWW]

please apply patch from this post

R36.3 Patch to re-enable GICv2m for PCIe MSI interrupts and restore I/O performance

R36.x has a major performance regression over R35.x. The problem is all PCIe MSI interrupts are now routed to CPU 0 whereas they used to be distributed across all CPU cores. For any workload requiring high I/O bandwidth, e.g. multiple NVME devices, this results in lower performance and system lock ups due to CPU 0 being overwhelmed with interrupts. There was an earlier thread on this issue but with no solution: [Orin NX R36.2 cannot change IRQ smp affinity - #7 by thegtx25] I’ve ported the o…

[3285195874]

thanks,
I’m able to adjust the interrupt affinity for this PCIe NIC, yet the aggregate throughput tops out at ~1.81 Gbps and fails to reach the full 2.5 Gbps line rate.