Other Linux Distributions Begin Analyzing Clear Linux's Performance Optimizations

Written by Michael Larabel in Clear Linux on 24 April 2016 at 09:55 AM EDT. 62 Comments
CLEAR LINUX
Fedora developers appear to be among those analyzing Intel's Clear Linux distribution for the performance optimizations made.

In case you haven't been following our many original performance tests in the past months, Intel's Clear Linux is surprisingly fast. Clear Linux just isn't fast for server/workstation/enterprise workloads but even for Intel graphics performance and other general workloads.

Clear Linux utilizes AutoFDO when building their packages, enables numerous CFLAGS/CXXFLAGS by default, does a lot of performance tuning, carries many patches to their Linux kernel, etc. They're also quick to jump on other possible avenues for performance optimizations like the recent talk about the Linux scheduler's effectiveness.

An interesting thread was started this week by Fedora developers about Intel's Clear Linux optimizations and what could be carried in the Fedora space. From initial indications, not too much. Beyond Clear Linux just focusing on Intel hardware, they optimize their software for just the recent generations of Intel hardware (Update: To clarify, Westmere and newer CPUs are supported). Fedora and other general purpose Linux distributions, of course, don't have that luxury.

Fedora kernel maintainer Josh Boyer also commented, "Specifically for the kernel, I believe Intel is carrying patches to enable LTO that are not upstream yet. Andi Kleen is working on pushing them upstream but they haven't made it as far as I know."

Richard Jones of Red Hat's Visualization Group had these interesting remarks to say:
The article you posted seems to concentrate on graphics performance which I've not looked at. However I have looked at Intel's boot / container performance in great detail.

I'm doing a bit of work improving the performance of qemu and the kernel to match what Intel Clear Linux can do (which is impressive, especially for running containers securely). Unfortunately the Intel kernel is heavily patched with non-upstream patches, and not even patches which could become upstream (they do things like commenting out chunks of code with single line commit messages). There's not really any way we can use the Intel work directly. Also Intel's kernel uses a very cut-down configuration, but Fedora kernel developers would prefer (with good reason) to ship a single kernel image for all use cases, and by necessity that is a "full fat" kernel.

So it's a long process. I've proposed a talk about this subject at the KVM Forum 2016 at the end of August.

Benchmarking Clear Linux's QEMU is something I haven't done yet, but now it certainly makes me want to put that on my TODO list for doing ASAP with the praise it's getting by Richard Jones.

Hopefully more Linux distributions will find ways to work together for improving the overall performance of Linux and seeing many of the relevant changes work their way back upstream.
Related News
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.

Popular News This Week