Announcement

Collapse
No announcement yet.

Looking At GNU/Linux's Performance Over 2016 With Intel's Clear Linux

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Looking At GNU/Linux's Performance Over 2016 With Intel's Clear Linux

    Phoronix: Looking At GNU/Linux's Performance Over 2016 With Intel's Clear Linux

    If you have been curious how the performance of the GNU/Linux stack has evolved over 2016, I ran some benchmarks of the rolling-release Clear Linux from the start of 2016 compared to this week to see how gains in the upstream software have evolved as well as their aggressive out-of-the-box optimizations for this operating system out of Intel's Open-Source Technology Center.

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Impressive.
    ## VGA ##
    AMD: X1950XTX, HD3870, HD5870
    Intel: GMA45, HD3000 (Core i5 2500K)

    Comment


    • #3
      I wonder if the same results or better can be achieved at Gentoo with best fine tuning possible, then make a tutorial about it

      Comment


      • #4
        Originally posted by timofonic View Post
        I wonder if the same results or better can be achieved at Gentoo with best fine tuning possible, then make a tutorial about it
        Probably.
        The real question is: are you willing to go through all the effort?

        Comment


        • #5
          Originally posted by nomadewolf View Post

          Probably.
          The real question is: are you willing to go through all the effort?
          Only if there's a full guide about all the process, then it could be automated and wait to compile (maybe I can get CPU cycles from others if possible)

          Comment


          • #6
            Originally posted by nomadewolf View Post

            Probably.
            The real question is: are you willing to go through all the effort?
            Yes, of course. Many Gentoo users do just that, and it's much easier since per package env files became a thing. Wins can be even higher for AMD systems and especially (in-order execution) Intel Atoms.

            Comment


            • #7
              Originally posted by timofonic View Post
              I wonder if the same results or better can be achieved at Gentoo with best fine tuning possible, then make a tutorial about it
              Yeah, it would be great if other distros and/or upstream would pick up some of these optimizations. They are meant to learn from each other, aren't they?

              BTW, Michael, some tests have both -O3 and -O2 (and some have neither). For the mentioned upcoming distribution comparison, I hope that you also run tests with the tests themselves compiled with the same compiler optimizations. It is definitely interesting to see the effect of different default compiler settings, but I think they shouldn't dominate the whole comparison. I'm used to a test trying to keep all variables constant except the thing being tested. I'm new to Linux, but don't applications usually define their own compiler settings? Are the compiler settings used to compile the kernel and the packages not different than the default "end user" compiler settings, with the "end users" usually modifying them for their own purpose?

              Comment


              • #8
                Originally posted by indepe View Post

                Yeah, it would be great if other distros and/or upstream would pick up some of these optimizations. They are meant to learn from each other, aren't they?

                BTW, Michael, some tests have both -O3 and -O2 (and some have neither). For the mentioned upcoming distribution comparison, I hope that you also run tests with the tests themselves compiled with the same compiler optimizations. It is definitely interesting to see the effect of different default compiler settings, but I think they shouldn't dominate the whole comparison. I'm used to a test trying to keep all variables constant except the thing being tested. I'm new to Linux, but don't applications usually define their own compiler settings? Are the compiler settings used to compile the kernel and the packages not different than the default "end user" compiler settings, with the "end users" usually modifying them for their own purpose?
                This is a function of the distribution. The distribution changes its compiler flags over time with continued research and performance data.

                Comment


                • #9
                  Originally posted by LinuxID10T View Post

                  This is a function of the distribution. The distribution changes its compiler flags over time with continued research and performance data.
                  Well, that much I know already. I think. When you talk about "default compiler flags", those are the ones that you get when you type "gcc helloWorld.c", right ?

                  And/or are those the same ones that a distribution uses to compile the kernel and the packages? And/or just the ones that Michael uses to compile the test programs?

                  My impression is that the test programs are compiled differently when running them on different distributions, but I'd (also) want to know if a (test-) application that's compiled with its own settings, how that will run on different distributions. Let's say it is always compiled with -O3, LTO, -m64, etc, for example, on each distribution.

                  Comment


                  • #10
                    I/o results and some synthetic benchmarks usually vary widely. The rest would probably perform identically as long as the compiler flags are the same.

                    Comment

                    Working...
                    X