Announcement

Collapse
No announcement yet.

Fedora 19 vs. Fedora 20 Linux Benchmarks

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

  • Fedora 19 vs. Fedora 20 Linux Benchmarks

    Phoronix: Fedora 19 vs. Fedora 20 Linux Benchmarks

    Published today are benchmarks from two Intel systems comparing the performance of Fedora 19 "Schr?dinger's Cat" to Fedora 20 "Heisenbug" for various workloads. Especially for those using open-source graphics drivers, Fedora 20 can be worth the upgrade for performance reasons.

    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
    Can anyone tell me, from the top of his head is fine, how much performance will I lose in Team Fortress 2, moving from legacy Catalyst 13.1 on Fedora 18 to Open source radeon drivers on Fedora 20? Other games? I'd be obliged.

    My hw is Radeon HD 3650 / C2D 2.6 / 8 GB RAM
    Last edited by Bucic; 20 December 2013, 09:42 AM.

    Comment


    • #3
      Originally posted by Bucic View Post
      from the top of his head is fine
      Great, then you'd lose about 20-50% performance depending on the game. TF2 in particular I have no idea. Not exactly a 3xxx generation gpu, but could still be relevant: http://www.phoronix.com/scan.php?pag...apu_1310&num=1

      Comment


      • #4
        Solid Release a bit boring.
        Whats nice is xbmc 13 alpha is included ^^.

        I am using it here now a while but use here i3-wm so I dont see much improvement, but at least nothing did broke
        But I am very very persuated about fedora today.


        I switched from Ubuntu Machines to Fedora and one archlinux machine now for around 6 Months or so. I was a bit 50/50 about what I like more Arch or Fedora.

        I like both much, both use systemd both have imho better default settings than ubuntu.

        1. ramfs for /tmp
        2. old kernels gets at some point deinstalled
        3. deliver a full current gnome-release
        4. no keylogger+trojan pre-installed
        ...

        but there were some stuff I liked more about fedora and stuff I liked more on arch side.

        fedora:

        - had faster mesa 9.2 support
        - gnome feeled better configured / faster
        - more focus on free software

        arch:

        - had xbmc 13 in AUR
        - better wiki
        - community driven


        so technicaly basicly the only good point for arch was its ability to install xbmc 13. Now and I dont know how long this package was in rawhide there is a xbmc package installable in F20, and in reality compiling of the xbmc 13 in arch did not work on my machine with arch a few days ago, I never tested it because I used this pc on another monitor, maybe it was because of the ramfs and only having 4gb ram I dont know but even if it would have sucedded it would take several hours on this often low power xbmc machines. (in this case a zacate), so the only kind of advantage is gone.

        Rolling release is also no argument first u may like it more or not, but even if u like it more arch has a rolling-release-branch so u can get that there too.

        So the better wiki of arch is the only good thing, but u can use 90-99% of the stuff in it directly in fedora. They are pretty similar except their packaging system.
        Last edited by blackiwid; 20 December 2013, 11:02 AM.

        Comment


        • #5
          Originally posted by Bucic View Post
          Can anyone tell me, from the top of his head is fine, how much performance will I lose in Team Fortress 2, moving from legacy Catalyst 13.1 on Fedora 18 to Open source radeon drivers on Fedora 20? Other games? I'd be obliged.

          My hw is Radeon HD 3650 / C2D 2.6 / 8 GB RAM
          It's hard to tell. Maybe you can find some older Phoronix benchmarks for your hardware. Or boot the live usb, mount /home from your disk and try to run your games. It should work I think. It would be good to install libtxc_dxtn from rpmfusion before tests. With 8GB RAM it should be doable on a live usb.
          Code:
          yum install http://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-20.noarch.rpm
          yum install libtxc_dxtn.{i686,x86_64} mesa-lib{GL,GLU,EGL,GLES}.i686

          Comment


          • #6
            Nice to see the latest generation Intel GT1 beating the last generation GT1, but it also raises one of the biggest problems with Intel GPUs. Why can't they just unlock the best GPU in all CPU models? Why unlock the best GPUs only on the best (more expensive) CPUs which are more likely to be used with discrete graphics cards anyway?

            Comment


            • #7
              devius, kruger,

              Thank you guys! 20-50% seems probable to me, after I read some more on the subject. Before, I was hoping a 25% drop would be the worst case. No, I cannot afford losing this much FPS. It's just sad that there are still trolls popping up here and there, yelling 'Using Catalyst doesn't make sense now!'. They need a reality slap.

              Re, testing it myself. I forgot my main Fedora install (I use another one, LXDE-based, with downgraded Xorg just for gaming) is not Catalysted (installing Catalyst still brakes booting with Intel graphics!) and I can simply:
              - Install Steam.
              - Install Portal 1 or Team Fortress 2 and point Steam to their existing install directories (coming from the LXDE Fedora)
              - test how Radeon drivers handle these games
              ...
              Ah, just recalled it may not be good for me to test gaming performance on Fedora with Gnome. Gnome yields worse performance by itself, topped with the tearing bug (on Intel GPUs)



              BTW, sorry for the funky grammar in my first post here. It was too late to edit it.

              PS. When I tried to complete at least some of the tests during the Fedora testing day, everything I tried ran faster on my integrated vintage GM45 Intel compared to my Radeon HD3650.
              Last edited by Bucic; 20 December 2013, 02:58 PM.

              Comment


              • #8
                Originally posted by Bucic View Post
                devius, kruger,

                Thank you guys! 20-50% seems probable to me, after I read some more on the subject. Before, I was hoping a 25% drop would be the worst case. No, I cannot afford losing this much FPS. It's just sad that there are still trolls popping up here and there, yelling 'Using Catalyst doesn't make sense now!'. They need a reality slap.
                I'd be surprised if it's over 25% on Mesa 10 and kernel 3.11+. If you're on 9.2, I don't know.

                Also, that's assuming it all works correctly. The r600 cards are not very well tested, and there have been regressions for those cards before that weren't caught very quickly.

                Comment


                • #9
                  Originally posted by smitty3268 View Post
                  I'd be surprised if it's over 25% on Mesa 10 and kernel 3.11+. If you're on 9.2, I don't know.

                  Also, that's assuming it all works correctly. The r600 cards are not very well tested, and there have been regressions for those cards before that weren't caught very quickly.
                  Ah, yes, I forgot to specify the stack - let's assume Mesa 10 and Kernel 3.11+.
                  I understand what you're saying here. Having an old HW is bad in two ways. First - it's slow, second - it doesn't get much developers' attention, as you said.

                  Well, my future rig will be Intel only and most probably an ITX factor.

                  Comment


                  • #10
                    F19 as initially released?

                    I probably missed it, but did the article mention at what point 'F19' was sampled? From the kernel version, it looks like it was F19 when it first released, not F19 with all current updates.

                    I guess that's useful from a "what have we gained in the past year or so" viewpoint, but it doesn't tell someone with an up to date F19 system what advantage there might be to upgrade to F20. The latter seems more useful.

                    Comment

                    Working...
                    X