Announcement

Collapse
No announcement yet.

Radeon Gallium3D Moved Closer To Performance Parity With AMD's Catalyst In 2014

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

  • Radeon Gallium3D Moved Closer To Performance Parity With AMD's Catalyst In 2014

    Phoronix: Radeon Gallium3D Moved Closer To Performance Parity With AMD's Catalyst In 2014

    Last week I wrote about the incredible improvements to AMD's open-source Linux driver over the course of 2014 that showed many significant OpenGL performance improvements for the open-source driver on various Radeon GPUs... But how does the latest open-source code compare to the closed-source Catalyst driver? In this article are benchmarks from an even larger assortment of Radeon GPUs while testing the latest Radeon Gallium3D and Catalyst drivers at the end of 2014.

    http://www.phoronix.com/vr.php?view=21284

  • #2
    I understand full well that the point of benchmarking is to check performance of applications that you actually use at settings you actually use. That's why I don't understand why you continuously benchmark at settings that result in hundreds of frames per second. Nobody is ever going to use those settings. Those benches don't mean anything at all. At those framerates you arent even benchmarking the GPU at all. If you really want to understand the performance of a GPU, then you need to recreate GPU bound conditions... But you don't even try...

    All you'd have to do is post the settings you used to achieve the benchmark and then even those would be repeatable. But you don't try that either.

    You can't make the the claim that performance is catching up, the benches you ran can't show that at those framerates.

    You want people to support this site financially, but you make no attempt at all to post game benchmarks that represent real world usage. There isnt any value in the benches you posted.....

    You can make worthwhile benches that are repeatable too.

    Comment


    • #3
      Plus you should check the cases that Catalyst loses by far and the AI that helps Catalyst cheat all famous benchmarks doesn't help, like Counter_Strike and Team_Fortress.

      Comment


      • #4
        Originally posted by duby229 View Post
        I understand full well that the point of benchmarking is to check performance of applications that you actually use at settings you actually use. That's why I don't understand why you continuously benchmark at settings that result in hundreds of frames per second. Nobody is ever going to use those settings. Those benches don't mean anything at all. At those framerates you arent even benchmarking the GPU at all. If you really want to understand the performance of a GPU, then you need to recreate GPU bound conditions... But you don't even try...

        All you'd have to do is post the settings you used to achieve the benchmark and then even those would be repeatable. But you don't try that either.

        You can't make the the claim that performance is catching up, the benches you ran can't show that at those framerates.

        You want people to support this site financially, but you make no attempt at all to post game benchmarks that represent real world usage. There isnt any value in the benches you posted.....

        You can make worthwhile benches that are repeatable too.
        The tests used were ones that (0. can be automated well), 1. work on both the open and closed Linux drivers and 3. also work fine under Windows under a similar quality and can be widely reproduced.

        With regard to the frame-rate, I could easily test at 4K as done often in articles or even just 2560x1600, but then I receive complaints that those resolutions are too uncommon and should test at more normal resolutions (e.g. 1080p). When I do tests at multiple resolutions, then I get complaints the articles are too long....
        Michael Larabel
        http://www.michaellarabel.com/

        Comment


        • #5
          I understand, But AA and AF can help bind up a GPU too. Resolution isn't the only way to load down a graphics processor. Additionally a bunch of games can be configured through in game settings. Many linux native games even have command line switches to automate in game settings.

          All I'm saying is that at such high framerates the GPU is not the bottleneck. You arent seeing GPU performance.

          And if all else fails and the games you are currently benchmarking can't grind down the GPU, then you really need to re-evaluate the games that you're benching.

          EDIT: Not to mention that the whole point of benching is to test games that are actually widely played.
          Last edited by duby229; 12-29-2014, 08:38 PM.

          Comment


          • #6
            When it comes to older cards like the 6k series, it is actually way beyond catalyst. You see, raw fps might be a very little bit higher with fglrx, but games played on fglrx microstutter like hell and have serious memory fragmentation issues even if you have 300 fps. Radeon might only have 290 fps, but it's smooth as silk.

            Comment


            • #7
              Originally posted by Michael View Post
              The tests used were ones that (0. can be automated well), 1. work on both the open and closed Linux drivers and 3. also work fine under Windows under a similar quality and can be widely reproduced.

              With regard to the frame-rate, I could easily test at 4K as done often in articles or even just 2560x1600, but then I receive complaints that those resolutions are too uncommon and should test at more normal resolutions (e.g. 1080p). When I do tests at multiple resolutions, then I get complaints the articles are too long....
              I wonder if you'd get complaints about making too many articles if you did multiple resos and posted them in multiple articles simultaneously

              Comment


              • #8
                I do agree that you only need to bench at one or two resolutions. 1080p and 4k would be fine with me.

                EDIT: except in cases like 6450 where framerates are too low, then a lower res bench would be ok too. Matter of fact I would take the 6450 out of the bench kit so that wouldnt be a problem.
                Last edited by duby229; 12-29-2014, 08:45 PM.

                Comment


                • #9
                  I'd almost be willing to bet that if you'd throw in some top tier linux titles into your suite, you'd get more hits on your articles.

                  Thing is there are fanboys of all types. I'm not saying that this site needs more fanboys, just saying fanboys do a lot of clicking.

                  Comment


                  • #10
                    Originally posted by Michael View Post
                    The tests used were ones that (0. can be automated well), 1. work on both the open and closed Linux drivers and 3. also work fine under Windows under a similar quality and can be widely reproduced.

                    With regard to the frame-rate, I could easily test at 4K as done often in articles or even just 2560x1600, but then I receive complaints that those resolutions are too uncommon and should test at more normal resolutions (e.g. 1080p). When I do tests at multiple resolutions, then I get complaints the articles are too long....
                    The bottom line is that you are heavily limited by the benchmarks you have to test.

                    1080p is a good test resolution, i wouldn't change that. (Well, unless you add in the low-level GPUs, like that 6450 - you should probably just drop it to 720p for all the tests, since it can't keep up with the other hardware).

                    The problem is that most readers here don't care about 2 of the 3 points above (automation, and equivalent windows support) and that heavily limits what you show. I can understand why you care about that, given your goals with Phoromatic, but there is definitely a conflict between what you want to show and what most of the readers here want to see. (Which is the most common games they actually play, which actually can stress the hardware, perhaps mixed in with a few tests like Unigine or artificial GPUTest results.)
                    Last edited by smitty3268; 12-29-2014, 08:56 PM.

                    Comment

                    Working...
                    X