Intel® Ethernet 700 Series

Linux Performance Tuning Guide

ID Date Version Classification
334019 12/09/2024 1.1 Public
Document Table of Contents
Document Table of Contents
Notices and Disclaimers Revision History ​Introduction Initial Checklist ​Baseline Performance Measurements and Tuning Methodology Tuning i40e Driver Settings Platform Tuning (i40e Non-Specific) ​Adapter Bonding Performance Troubleshooting Recommendations for Common Performance Scenarios
Notices and Disclaimers
Revision History
​Introduction
​Introduction ​Related References
​Related References
Initial Checklist
Update Driver/Firmware Versions ​ Read the README ​Check That Your PCI Express (PCIe) Slot Is x8 ​Check System Hardware Capabilities ​Ensure DDP Package is Loading Properly
Update Driver/Firmware Versions
​ Read the README
​Check That Your PCI Express (PCIe) Slot Is x8
​Check System Hardware Capabilities
​Check System Hardware Capabilities ​ Kernel Boot Parameters
​ Kernel Boot Parameters
​Ensure DDP Package is Loading Properly
​Baseline Performance Measurements and Tuning Methodology
​Network Performance Benchmarks ​Tuning Methodology
​Network Performance Benchmarks
​Network Performance Benchmarks ​iPerf2 ​iPerf3 ​netperf
​iPerf2
​iPerf3
​netperf
​Tuning Methodology
Tuning i40e Driver Settings
​IRQ Affinity ​Tx/Rx Queues ​Interrupt Moderation ​Ring Size ​Flow Control ​Jumbo Frames
​IRQ Affinity
​Tx/Rx Queues
​Interrupt Moderation
​Ring Size
​Flow Control
​Jumbo Frames
Platform Tuning (i40e Non-Specific)
​BIOS Settings ​Power Management Intel® Turbo Boost ​Firewalls ​Application Settings ​Kernel Version ​Operating System/Kernel Settings Network Device Backlog Platform-Specific Configurations and Tuning
​BIOS Settings
​Power Management
​Power Management ​C-State Control ​PCIe Power Management ​CPU Frequency Scaling Additional Power Management Guidance
​C-State Control
​PCIe Power Management
​CPU Frequency Scaling
Additional Power Management Guidance
Intel® Turbo Boost
​Firewalls
​Application Settings
​Kernel Version
​Operating System/Kernel Settings
Network Device Backlog
Platform-Specific Configurations and Tuning
Platform-Specific Configurations and Tuning 4th Generation Intel® Xeon® Scalable Processors AMD EPYC
4th Generation Intel® Xeon® Scalable Processors
AMD EPYC
​Adapter Bonding
Performance Troubleshooting
​CPU Utilization ​i40e Counters ​Network Counters ​System Logs ​Intel svr-info Tool
​CPU Utilization
​i40e Counters
​Network Counters
​System Logs
​Intel svr-info Tool
Recommendations for Common Performance Scenarios
​IP Forwarding ​Low Latency
​IP Forwarding
​Low Latency

​Tx/Rx Queues

The default number of queues enabled for each Ethernet port by the driver at initialization is equal to the total number of CPUs available in the platform. This works well for many platforms and workload configurations. However, in platforms with high core counts and/or high Ethernet port density, this configuration can cause resource contention. Therefore, it might be necessary in some cases to modify the default for each port in the system.

The default number of Tx/Rx queues can vary depending on the specific model and driver version. The number of queues can be adjusted using the ethtool -L command listed below.

Note:In these cases, Intel recommends that you reduce the default queue count for each port to no more than the number of CPUs available in the NUMA node local to the adapter port. In some cases, when attempting to balance resources on high port count implementations, it might be necessary to reduce this number even further.

To modify queue configuration:

The following example sets the port to 32 Tx/Rx queues:

ethtool -L ethX combined 32

Example output:

ethtool -l ethX Channel parameters for ethX: Pre-set maximums: RX: 96 TX: 96 Other: 1 Combined: 96 Current hardware settings: RX: 0 TX: 0 Other: 1 Combined: 32
​Interrupt Moderation