Multi-Core, Multi-OS Scaling Performance

Written by Michael Larabel in Operating Systems on 21 February 2011 at 01:09 AM EST. Page 6 of 6. 28 Comments.

The last test today is MG.B. There isn't anything too new here: Fedora 14 scaled very well with the increasing core count, Hyper Threading was of limited benefit, and PC-BSD continued to falter at three and four cores.

This is the first time publishing test results on Phoronix that look at the multi-core performance across multiple-operating systems while using a common compiler and also normalizing the results against their single-threaded values. These results illustrate that overall the Linux, PC-BSD (FreeBSD), and OpenSolaris operating systems scale quite similar up through at least six cores.

Fedora 14 with the Linux 2.6.35 kernel did exhibit stronger relative performance in some test profiles when running above two cores, but in other test profiles, there was no difference. When looking at the impact of Hyper Threading, the OpenSolaris-based OpenIndiana did better with Hyper Threading enabled compared to the other three operating systems, but in some tests (NPB) all four operating systems were actually impaired by this Intel technology. The only other odd finding with these results was how PC-BSD/FreeBSD had not scaled as well as the other operating systems when using three and four cores, but with two and six cores, it matched the other operating systems.

For those wanting to see how the performance breaks down overall, here is the OpenBenchmarking.org-generated overview showing the geometric mean of all the tests that were part of the test queue.

As the results up to now have been normalized against their single core values, for those wondering what operating system had the best performance overall rather than just scaling the best, it was Fedora 14. Via OpenBenchmarking.org it was also just a matter of a few clicks to take all of the test results, transpose the comparison to compare the operating systems rather than core count, group the values by operating system, and then generate a graph showing the harmonic mean.

Fedora 14 was distinguishably different from the rest while CentOS 5.5 and PC-BSD 8.1 were close while OpenIndiana b148 was a bit further out. OpenBenchmarking.org will launch at the Southern California Linux Expo with Phoronix Test Suite 3.0, after which you are able to examine and further manipulate this data-set (and any other data) yourself.

If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.


Related Articles
About The Author
Michael Larabel

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 20,000 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, LinkedIn, or contacted via MichaelLarabel.com.