Intel® Ethernet 800 Series

Linux Performance Tuning Guide

ID Date Version Classification
636781 07/19/2025 Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents
Document Table of Contents
Notices and Disclaimers Revision History Introduction Initial Checklist ​Baseline Performance Measurements and Tuning Methodology Performance Troubleshooting ​Tuning ice Driver Settings ​Platform Tuning (ice Non-Specific) ​Operating System Tuning ​Recommendations for Common Performance Scenarios
Notices and Disclaimers
Revision History
Introduction
Introduction ​Terminology ​Reference Documents
​Terminology
​Reference Documents
Initial Checklist
​Update Driver/Firmware Versions ​Read the ice Driver README ​Check System Hardware Capabilities ​Check Linux Kernel Version ​Check Kernel Boot Parameters ​Check System Logs for Driver Errors ​Check PCI Express (PCIe) Slot Capabilities ​Ensure DDP Package is Loading Properly
​Update Driver/Firmware Versions
​Read the ice Driver README
​Check System Hardware Capabilities
​Check Linux Kernel Version
​Check Kernel Boot Parameters
​Check System Logs for Driver Errors
​Check PCI Express (PCIe) Slot Capabilities
​Ensure DDP Package is Loading Properly
​Baseline Performance Measurements and Tuning Methodology
​General Performance Tuning Methodology ​Network Performance Benchmarks
​General Performance Tuning Methodology
​Network Performance Benchmarks
​Network Performance Benchmarks ​iperf2 ​iperf3 ​netperf ​Adapter Bonding
​iperf2
​iperf3
​netperf
​Adapter Bonding
Performance Troubleshooting
​CPU Utilization ​ice Counters ​Intel Unified Driver Statistics ​Intel PerfSpect Tool ​IOMMU in Virtualized Environments ​Resource Limitations Using E810 in 8x10G Mode
​CPU Utilization
​ice Counters
​Intel Unified Driver Statistics
​Intel PerfSpect Tool
​IOMMU in Virtualized Environments
​Resource Limitations Using E810 in 8x10G Mode
​Tuning ice Driver Settings
​IRQ Affinity ​Tx/Rx Queues ​Transmit and Receive Queue Alignment ​Transmit Scheduling Layers ​Interrupt Moderation ​Ring Size ​Flow Control ​Jumbo Frames ​Accelerated Receive Flow Steering (aRFS)
​IRQ Affinity
​Tx/Rx Queues
​Transmit and Receive Queue Alignment
​Transmit Scheduling Layers
​Interrupt Moderation
​Ring Size
​Flow Control
​Jumbo Frames
​Accelerated Receive Flow Steering (aRFS)
​Platform Tuning (ice Non-Specific)
​BIOS Settings ​Platform Power Management ​CPU Frequency Scaling Intel® Turbo Boost ​Additional Power Management Guidance Transmit(Tx) and Receive(Rx) buffer size Network Device Backlog Platform-Specific Configurations and Tuning
​BIOS Settings
​Platform Power Management
​Platform Power Management ​CPU Package C-State Control
​CPU Package C-State Control
​CPU Frequency Scaling
Intel® Turbo Boost
​Additional Power Management Guidance
Transmit(Tx) and Receive(Rx) buffer size
Network Device Backlog
Platform-Specific Configurations and Tuning
Platform-Specific Configurations and Tuning 4th Gen Intel® Xeon® Scalable Processors AMD EPYC
4th Gen Intel® Xeon® Scalable Processors
AMD EPYC
​Operating System Tuning
​Firewalls ​Application Settings ​Kernel Version ​Kernel Settings
​Firewalls
​Application Settings
​Kernel Version
​Kernel Settings
​Recommendations for Common Performance Scenarios
​IP Forwarding ​Low Latency ​Related Performance Topics
​IP Forwarding
​Low Latency
​Related Performance Topics
​DPDK ​ADQ
​DPDK
​ADQ

​Transmit and Receive Queue Alignment

Having Tx/Rx queue alignment can be helpful in some configurations such as some multi-threaded workloads or busy polling environments where it is not possible to pin the threads to a CPU. The set_​xps_​rxqs script is designed to align the Tx queue based on the Rx queue(s) map. This is done by configuring Rx queue(s) map per Tx queue via sysfs. This Rx queue(s) map is used during selection of Tx queue in the data path.

The performance benefit of aligning Tx/Rx queues varies between different workloads.

Using the set_​xps_​rxqs script from the ice source package:

<path-to-ice-package>/scripts/set_xps_rxqs <ethX>
​Transmit Scheduling Layers