Intel® Ethernet 600 Series

Linux Performance Tuning Guide
Ethernet Products Group (EPG)

ID Date Version Classification
861164 08/28/2026 Public
Document Table of Contents
Document Table of Contents
Notices and Disclaimers Revision History ​Introduction ​Initial Checklist ​Baseline Performance Measurements and Tuning Methodology Platform Tuning (ixgbe Non-Specific) Changing Link Speed Adapter Bonding Performance Troubleshooting
Notices and Disclaimers
Revision History
​Introduction
​Introduction ​Related References Compatibility Notes Known Issues
​Related References
Compatibility Notes
Known Issues
​Initial Checklist
Update Driver/Firmware Versions ​Read the README Check PCI Express (PCIe) Slot Width ​Check System Hardware Capabilities
Update Driver/Firmware Versions
​Read the README
Check PCI Express (PCIe) Slot Width
​Check System Hardware Capabilities
​Baseline Performance Measurements and Tuning Methodology
​Network Performance Benchmarks
​Network Performance Benchmarks
​Network Performance Benchmarks ​iPerf2 ​iPerf3 ​netperf
​iPerf2
​iPerf3
​netperf
Platform Tuning (ixgbe Non-Specific)
​Tuning Methodology ​BIOS Settings ​Firewalls
​Tuning Methodology
​BIOS Settings
​Firewalls
Changing Link Speed
Adapter Bonding
Adapter Bonding Configuration Using Network Configuration Files
Configuration Using Network Configuration Files
Performance Troubleshooting
​CPU Utilization ixgbe Counters ​Network Counters ​System Logs ​Intel PerfSpect Tool
​CPU Utilization
ixgbe Counters
​Network Counters
​System Logs
​Intel PerfSpect Tool

​iPerf2

Intel recommends iperf2 over iperf3 for most throughput benchmarking because it supports true multi-threaded traffic generation within a single instance, enabling efficient utilization of multiple CPU cores, Receive Side Scaling (RSS) queues, and hardware queues. This improves scalability and allows higher, more consistent throughput on high-speed Ethernet links compared to iperf3, which often requires multiple application instances to achieve similar performance.

For 600 Series devices, one thread should achieve line rate at 100 Mbps. Use up to 4 threads with the -P option to achieve 10 Gbps.

  • To run unidirectional traffic from client to server:

    Server command example:

    iperf2 -s

    Client command example:

    iperf2 -c <serverIP> -P <threads>
  • To run bidirectional traffic from client to server and vice versa:

    Server command example:

    iperf2 -s -p <port>

    Client command example:

    iperf2 -c <serverIP> -p <port> -P <threads> --full-duplex

    or

    iperf2 -c <serverIP> -p <port> -P <threads> -d
Note:​Both the --full-duplex and -d options in iperf2 allow users to perform bidirectional testing. With --full-duplex traffic runs in both directions at the same time. With -d traffic runs in both directions sequentially rather than simultaneously. Note:​When testing iperf2 across multiple server ports, the -D flag can be added to the server command to run all server sessions in the background from the same terminal window. The -D flag can also be used when the server command is embedded inside a for-loop in a script.
​iPerf3