Linux 3.17 To 4.15 Kernel Benchmarks On Intel Gulftown & Haswell
Written by Michael Larabel in Software on 26 January 2018. Page 1 of 1. 3 Comments

Here is a look at how the Linux kernel performance has evolved since the release of Linux 3.17 in October 2014. With all the major kernel releases over the past 3+ years, here is how the performance compares using two very different Intel Gulftown and Haswell systems.

I spent a good portion of the week running some Linux 4.0 to 4.15 benchmarks in large part motivated to see how the Meltdown and Spectre mitigation in Linux 4.15 has impacted the kernel performance in the grand scheme of things. I ended up taking all the kernel testing back to Linux 3.17: unfortunately that's where testing had to end since Linux 3.16 and prior wouldn't boot with the newer Linux user-space due using systemd. So on an Intel Core i7 990X and Core i7 5960X systems I compared the performance using each major Linux x.y.0 release from 3.17.0 to 4.15 Git. The lone exception is having to skip Linux 4.7 because when trying to boot that with EXT4 formatted from Ubuntu 18.04 daily, that kernel could only boot with the file-system in read-only mode.

If there is enough interest on some older systems I may do an even larger kernel comparison while using Ubuntu 12.04 LTS or 14.04 LTS user-space components in order to work on even older kernel releases. Obviously in order to go back to Linux 3.17, I had to use older system components as well for maintaining kernel/driver support. The two systems consisted of:

Core i7 990X - The high-end "Gulftown" processor from 2011. The Core i7 990X is a six-core processor plus Hyper Threading and has a 3.46GHz base clock frequency with 3.73GHz turbo frequency. The i7-990X system was paired with an MSI X58M motherboard, 3 x 4GB DDR3-1666 memory, and PNY CS1211 120GB SATA 3.0 SSD.

Core i7 5960X - The Haswell-E system providing a more recent look at Linux performance potential and as a refresher is an eight core part with Hyper Threading, 3.0GHz base frequency, and 3.5GHz turbo frequency. This system was paired with the ASRock X99 Extreme3 motherboard, 4 x 4GB DDR4-3000 memory, 120GB Intel SSDSC2BW12, and AMD FirePro V7900 graphics card.

On both systems was Ubuntu 18.04 LTS daily with an EXT4 file-system and GCC 7.2.0 compiler. Each major Linux kernel release was obtained from the Ubuntu Mainline Kernel PPA. These benchmarks were carried out using the Phoronix Test Suite.

Most of the kernels behaved relatively the same between the two Intel x86_64 systems, albeit on different scales. This testing isn't meant to compare i7-990X vs. i7-5960X but two different data points for looking at how the Linux kernel performance has changed over the past 3+ years since Linux 3.17's release in October 2014.

In the PostMark I/O benchmark that's primarily limited by fsync performance, Linux 4.15 indeed takes a significant drop in performance due to Kernel Page Table Isolation (KPTI) and Retpolines.

The ebizzy test, which aims to resemble web server workload performance, takes a much more noticeable impact on the more powerful Haswell-E Core i7-5960X system. But still Linux 4.15 lets its performance still run much better than the 3.xx or early 4.x kernel releases.

Interestingly the boot time on Linux 4.15 for the i7 990X Gulftown system is at its best while the i7-5960X system is still booting much slower than on pre-4.5 kernels.

GIMP is one of the real-world programs seeing a measurable slowdown in performance with Linux 4.15 due to KPTI/Retpoline.

The synthetic socket activity benchmark from Stress-NG also shows its performance hit on both systems from Spectre v2 and Meltdown mitigation.

The context switching performance hit is also evident.

On both systems, the Apache web server performance has gradually gotten slower with the newer kernel releases.

Hackbench is another kernel benchmark showing the performance impact on Linux 4.15.

Overall on both of these systems we find Linux 4.15 to take a performance hit in workloads dealing with a lot of kernel interactions, but at least in some instances the lower performance is still ahead of where the performance was on much older kernel releases. Thankfully though some performance optimizations are in the works that may hit for the upcoming Linux 4.16 merge window. Once the Linux 4.16 development begins to settle down, stay tuned for results being added in to these system metrics along with plenty of other benchmarks on Phoronix.

About The Author
Author picture

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 10,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter or contacted via MichaelLarabel.com.

Related Articles
Trending Linux News