Announcement

Collapse
No announcement yet.

Linus On GEM Patches: UNTESTED CRAP

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

  • #41
    Originally posted by josvazg View Post
    When something is new and unstable it usually is wise to keep the old stuff around so users can choose to fallback to it while the new stuff gets stable and has a good performance.

    Why didn't they choose this path?
    Why should we suffer poor linux drivers while Windows works as expected?

    Couldn't they keep the closed source versions around while the opensource ones mature?

    You DON'T close the old road till the new highway is ready.

    More examples?
    - AMD/ATI is keeping their closed source drivers while the opensource ones are being enhanced.
    - The OSS to ALSA path.

    It is painfully long, but the alternative is also quite long and in the meantime you have nothing working fine.

    It seems the trend for last little while is to push stuff out into the mainstream before it's really ready. KDE 4 and PulseAudio are a few other examples where it is made a "defacto standard" before they are really ready for prime time. It seems what we all have bitched about game devels pushing out crap before it's ready has now become a development model to follow for the rest of the industry. Release and fix later. It's really too bad, it hurts linux's rep for stability.
    Last edited by deanjo; 01 January 2009, 04:17 PM.

    Comment


    • #42
      Originally posted by josvazg View Post
      Intel's linux drivers performance has been decreasing from Ubuntu 7.10 (and distros of its same age) to now, even before the GEM phase.

      With Ubuntu 7.04's intel drivers, you could easily get 1500fps at glxgears, on Ubuntu 7.10 you got less that 1500fps, 8.04 lowers the mark to 1100fps, 8.10 to 500pfs and in current JAUNTY alphas TO 300fps!!
      60fps ought to be enough for anybody

      Seriously though I don't think glxgears is really useful for performance measurement. On my 780G IGP I "only" get between 900 and 1000fps with glxgears, but I do get playable frame rates with Half Life 2 Episode Two at 1280x800, and that's what matters. Not some stupid artificial "benchmark".

      Comment


      • #43
        Originally posted by monraaf View Post
        60fps ought to be enough for anybody

        Seriously though I don't think glxgears is really useful for performance measurement. On my 780G IGP I "only" get between 900 and 1000fps with glxgears, but I do get playable frame rates with Half Life 2 Episode Two at 1280x800, and that's what matters. Not some stupid artificial "benchmark".
        Yep. If memory serves, glxgears is (to the extent that it's a benchmark) just as much a benchmark of how fast the back buffer can be erased/flipped as it is a benchmark of how fast a 3D scene can be drawn, and as such is extremely sensitive to any kind of per-frame API overhead. In any case, it doesn't represent any kind of realistic application. Wasn't it necessary in some builds (not sure if it was stock for some period or a distro addition) to add a flag like --i-understand-that-this-program-is-not-a-benchmark in order to enable the fps output?

        Comment


        • #44
          That's the big problem with glxgears; because it only exercises the most basic functions, older chips tend to run faster than newer chips (old chips spent their transistors on ROPs, newer chips spend their transistors on shader processors) and heavily optimized drivers sometimes run slower than simple drivers (if the shader compiler does more optimizing it speeds up complicated things but makes simple things run slower 'cause you often end up doing the optimizing even if there are no gains to be had).

          On the other hand, sometimes glxgears just runs slow because the new driver is crap
          Test signature

          Comment


          • #45
            Yes, for Xorg 7.1.1 glxgears needed an option, but -printfps did the same as -iacknowledgethatthistoolisnotabenchmark

            Comment


            • #46
              Originally posted by Kano View Post
              Yes, for Xorg 7.1.1 glxgears needed an option, but -printfps did the same as -iacknowledgethatthistoolisnotabenchmark
              A flag they should have never gotten rid of.

              Comment


              • #47
                Originally posted by monraaf View Post
                Seriously though I don't think glxgears is really useful for performance measurement.
                On my laptop with i915GM and on my netbook with i945GM i have always seen a strong relation between glxgears output and openarena benchmark.
                Maybe there are counterexamples, but i haven't found any.

                i get 1450 fps with ubuntu 8.04, and only 800 with ubuntu 8.10, and 3D effects, Google Earth, Openarena show the same gap between the 2 versions.

                Comment


                • #48
                  Originally posted by DanL View Post
                  I don't always agree with Linus' views, but can you blame him for blowing his lid when he has to repeatedly remind certain people of certain policies?
                  He knows he's being jerked around. If you can't improve windows break linux. I'm sure he's being as polite as he can possibly be but you pretty much have to use an enforcer on Intel and Microsoft.

                  Comment

                  Working...
                  X