Announcement

Collapse
No announcement yet.

9-Way Linux Desktop Distribution Benchmarks With The Intel Core i7 8086K

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

  • 9-Way Linux Desktop Distribution Benchmarks With The Intel Core i7 8086K

    Phoronix: 9-Way Linux Desktop Distribution Benchmarks With The Intel Core i7 8086K

    Chances are if you are spending more than $400 USD to have the Intel Core i7 8086K, the limited edition processor that is Intel's first to have a turbo frequency at 5.0GHz (and can easily overclock on all cores to 5.0+ GHz), you probably care a great deal about your system's performance. For squeezing extra performance out of the hardware, there is a wide variety of software optimizations available. Many of those software optimizations can be found within Intel's own Clear Linux distribution as previously shown while for this i7-8086K benchmarking is a look at how nine Linux distributions compare out-of-the-box when tested on this Coffeelake CPU and all CPU cores overclocked to 5.0GHz.

    http://www.phoronix.com/vr.php?view=26550

  • #2
    So Fedora on Weyland, any chance running it without it? Like lets say Mate DE?

    Comment


    • #3
      just a note for Debian stable, back ports has update Mesa, also a kernel and my guess would be intel gpu support. 4.16 is in backports, most of the time when something hits testing that is important it gets added to back ports, pretty fast. 4.17 is currently in unstable, soon as it hits testing it will get backported.Im shocked that the by pass security update was not backported to the old stable kernel, since i trust debian I assume it has not been fixed because the whole thing is a joke. Like the huge security whole in Ryzen that could only be exploited if you were logged in as root, Linus went off on that.Really if you are logged in as root you can install dirty code, lol NO SHIT. Heartbleed was another one, where the world went nuts, so I went to the debian mail list to see what they said, as a note "no ssl private keys were exposed from debian" like they were on other distros and BSD. If debian gets excited i will ot
      I ran clear on my Ryzen box the brand new one posted in july, someone here ask had anyone tried to use it as a desktop, I planned to do just that but there were issues with things you just dont expect to have on linux in 2018. Like even firefox was broken on some websites, fedora had the same issue, unknown mime type unable to play the media, on both distros. Im sure its easy to fix, but in 2018, I dont expect that stuff. I expect everything to work out of the box. Out of all the distros I played with the past few weeks only Solus was worth the time. I have been testing with wine stagging and dxvk, blizzard games. Solus seems to handle deps really well, I did not have to hunt around to figure out what package did not get installed to make things work. Fedroa even made my old WD hard drive sound like it was coming apart quiet odd on Debian I dont even hear it and its super fast for a spinning drive. I ran windows 10 for about a year, a girl at my house had a laptop and 10 seemed pretty quick, the blizzard games I play were having issues with wine, over the past few months of updates on windows 10, on a 4790k or Ryzen 1800+ Debian boots faster from a spinning drive than windows does from a $400 m2 ssd. There have been some bios updates involved in the slowness it is possile that Asus changed something, to slow it down to make me buy an Asus super ssd. The ryzen 7 booted at 3.7 as a base clock on the original bios all default and now boots at 3.6 on the same defaults
      Last edited by Divergent1; 07-09-2018, 11:59 AM.

      Comment


      • #4
        Way to go, Debian!

        Comment


        • #5
          well... pretty good job ClearLinux. You give a performance boost which compares with a newer generation hardware purchase!

          Comment


          • #6
            Is that xfs having such an impact on tumbleweed?

            Comment


            • #7
              Why does Clear struggle with VP9 so much? What is it about that lib that makes it so hard for it to process?

              Comment


              • #8
                Typo:

                Originally posted by phoronix View Post
                currently using Linux 4.17, GNOME SHell 3.28.2, and GCC 8.1.
                (although GNOME *is* hell)

                Comment


                • #9
                  Originally posted by edwaleni View Post
                  Why does Clear struggle with VP9 so much? What is it about that lib that makes it so hard for it to process?
                  that is a very good question and you just made me go check

                  Comment


                  • #10
                    Originally posted by arjan_intel View Post

                    that is a very good question and you just made me go check
                    I didn't want to insinuate it performed poorly, hardly.

                    MP3 and Compression related tests show large levels of performance gains, but VP9 merely matches. That may not be a bad thing overall but in relative terms seems out of place.

                    I went back to look at Michael's change history on the test for a compare:

                    pts/vpxenc-2.1.1 [17 Jun 2018 11:20:20 EDT]
                    - Limit the thread count to 64, if above 64, vpxenc complains of CPU thread count being too high.

                    pts/vpxenc-2.1.0 [17 Jun 2018 10:58:27 EDT]
                    - Update build script, switch to new source file.

                    pts/vpxenc-2.0.0 [06 Feb 2018 11:39:48 EST]
                    - Update against libvpx 1.7.0, use new source video, various parameter updates, etc. Big rework overall.

                    and went through the historical test reports. but the CPU based encode rates were measured differently before this. That matches what his notes say.

                    I looked at the test script:

                    THREADCOUNT=$(($NUM_CPU_CORES>64?64:$NUM_CPU_CORES ))
                    LD_PRELOAD=../lib/libvpx.so.5 ./vpxenc --good --codec=vp9 -v --threads=\$THREADCOUNT --tile-columns=4 -o /dev/null ~/Jockey_1920x1080_120fps_420_8bit_YUV.y4m --width=1920 --height=1080 2> \$LOG_FILE

                    If I did my math right, that test file is a 3Gbps video. I would think with Clear's optimal use of CPU instructions and flags, it should have shown at least a 12-18% frame rate improvement over a stock OS. (somewhere near 21-23 fps)

                    As I said before, this isn't a complaint, just an observation that in most encode/decode tasks, Clear usually outperforms its peers, but in this case it matched.

                    Comment

                    Working...
                    X