Servers: The Right Compute
Showing results for 
Search instead for 
Did you mean: 

Reducing latency with HPE ProLiant Gen10 Servers


Tackling low-latency challenges? Check out our HPE white paper, updated for ProLiant Gen10 servers: Configuring and tuning HPE ProLiant Servers for low-latency applications.

Blog_Low_Latency.jpgWhen discussing performance with server solution architects, two metrics that most always seem to come up: throughput and latency.

Throughput is usually on the forefront of the vast majority of discussions around performance. The goal in most every design is to continuously increase overall compute throughput—faster transaction rates, faster number crunching and higher data retrieval rates are just a few examples of throughput. The desire to get more done in the least amount of time almost seems obvious.

In contrast, latency, while inherent in the computing environment, is something that designers are trying to diminish. Latency is the metric that describes how long your application has to wait before it can start doing meaningful work again. There are many sources of latency in an end-to-end compute solution and HPE offers resources to help tackle the problem.

Continued commitment to reducing latency

For several generations of ProLiant servers, HPE has published a technical whitepaper targeted at helping you tune your servers for low-latency. This comprehensive resource includes recommendations for hardware configurations, BIOS settings, and operating system settings that can result in decreased overall latency and increase performance in latency sensitive applications. We just updated the low-latency whitepaper again to include recommendations for our newest Gen10  server lineup.

Highlights from the updated version include:

  • Updated information specific to the new Intel Xeon Scalable Processor family
  • Leveraging Intelligent System Tuning for low-latency
  • Updated operating system tunings
  • Tuning with HPE iLO RESTful API and iLO 5
  • Updated guidance on System Management Interrupts

Reduction of System Management Interrupts (SMIs)

SMIs are often leveraged by system firmware for monitoring the health of the platform and/or manage power efficiency. Although SMI rates are typically kept low in order to avoid impact to overall throughput performance, they do introduce latency. When designing Gen10, HPE firmware engineering focused on ways to reduce the impact of SMIs for latency sensitive customers. They delivered in two meaningful ways. First, firmware based power management no longer makes use of periodic SMIs. For Gen10, the option to disable processor power and utilization monitoring has been deprecated as it doesn’t have an impact on latency. Second, the Memory Pre-Failure Notification feature no longer enables periodic SMIs until ECC errors are detected. This means that latency sensitive customers can now choose to keep Memory Pre-Failure Notifications enabled and only experience SMIs when correctable errors occur.

Intelligent System Tuning: new features that target low latency

HPE Intelligent System Tuning (IST) offers two new key features that target the low-latency market

  • Workload Matching—Quickly configure your server using workload based UEFI settings profiles which includes a profile targeted for latency-sensitive workloads.
  • Jitter Smoothing—Reduce the latency associated with enabling Turbo Boost mode while offering up to 12% increase(1) in frequency over base frequency

HPE is committed to helping you optimize you compute solutions for low-latency. You can find out more about it at the IST website which also conveniently has links to all the resources including the low-latency whitepaper, the IST whitepaper, tuning guides and more. Check it out:  

HPE Discover Madrid.jpeg

Attending the HPE Discover 2017 Madrid conference? Be sure to check out our session # B4561 on how you can Accelerate Performance with Intelligent System Tuning while you are there.





Scott Faasse
Master Technologist, Server Firmware and Performance
Hewlett Packard Enterprise

twitter.gif @HPE_Servers


(1) HPE Internal testing from Performance Engineering Benchmarking team, April 2017.

0 Kudos
About the Author


Scott Faasse is a Distinguished Technologist and HPE’s expert on Platform and Processor Power Management and Performance. Since joining Compaq/HP/HPE in 2001, Scott has served as the lead platform firmware developer for six generations of the ProLiant DL380 Server, architected and developed HPE’s Power Regulator feature, lead HPE in several industry standards and partner collaboration efforts, and is one of the principal technologist behind HPE’s Intelligent System Tuning. Scott is also an avid outdoorsman. He enjoys hiking long distances with a really heavy backpack (rucking), camping with family, fly fishing in urban settings, and traditional archery.