Announcement

Collapse
No announcement yet.

Mesa 9.1 Results Are Mixed For Radeon Gallium3D

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

  • Mesa 9.1 Results Are Mixed For Radeon Gallium3D

    Phoronix: Mesa 9.1 Results Are Mixed For Radeon Gallium3D

    After last week delivering benchmarks that showed Intel graphics being faster with Mesa 9.1 relative to earlier Mesa 3D releases, up today are benchmarks of Radeon Gallium3D (R600g) to compare the Mesa 9.1 performance to Mesa 9.0.2, 8.0.5, and 7.11.2.

    http://www.phoronix.com/vr.php?view=18445

  • #2
    Oh, what about more recent and powerful GPUs?

    You see, 4650 is both ancient and weak. How about benchmarking something more interesting like HD5000/6000 GPUs? Unfortunately, for some obscure reasons, most of phoronix benchmarks for Radeon are done on such an ancient GPU. You see, these days you can't buy HD4000 GPU in unused condition anyway.

    Furthermore, core i7 + 4650? Oh, that's a really decent example of very unbalanced system. Hi-end modern CPU + Low-end and outdated GPU? Really strange scenario.
    Last edited by 0xBADCODE; 02-06-2013, 12:51 PM.

    Comment


    • #3
      Hmm. I bet the reason why he uses a high end CPU when testing graphics cards is so that there won't be weird performance drops because of the cpu being under-powered.

      Comment


      • #4
        High-end cpu + low-end gpu is my setup exactly

        Comment


        • #5
          You see, 4650 is both ancient and weak. How about benchmarking something more interesting like HD5000/6000 GPUs? Unfortunately, for some obscure reasons, most of phoronix benchmarks for Radeon are done on such an ancient GPU. You see, these days you can't buy HD4000 GPU in unused condition anyway.
          Open source graphics drivers benchmarks for older series Radeon HD 4000, 3000, 2000 and even older are necessary. There are no alternative drivers for these cards after amd has dropped support.

          Comment


          • #6
            The 4650 is an older mid-range card, and not "ancient and weak" by any means. Radeon X300 = ancient and weak.

            Comment


            • #7
              Originally posted by DanL View Post
              The 4650 is an older mid-range card, and not "ancient and weak" by any means. Radeon X300 = ancient and weak.
              But still usable. My father-in-law is using my old Inspiron 6000 with a x300, and my parents were using an x600 in their desktop until a few weeks ago when the motherboard took a dump.

              And I agree that a 4650 is still perfectly usable. My work desktop has a core i7-2600k with 8GB RAM coupled with a radeon 5400, which is weaker than the 4650.

              Comment


              • #8
                As others have posted, it makes a lot of sense to keep benchmarking these cards as they are more likely to have people using them with the free radeon drivers than newer ones where Catalyst is still an option for people. Besides, I am still running quite happily with my 4670.

                As usual though, Larabel was light on details as to what could be slowing 9.1 down. There are vague references to it being because of the new AA code, but are new graphical features being enabled the only reason for the slowdowns? If one does not activate them, is the performance the same relative to other versions or even better (or still worse)? These are questions I would love to see explored, but I seriously doubt we will see them.

                Comment


                • #9
                  Originally posted by Hamish Wilson View Post
                  As usual though, Larabel was light on details as to what could be slowing 9.1 down.
                  No time/resources to do all that everytime myself, but anyone is welcome to reproduce the tests and explore more.
                  Michael Larabel
                  http://www.michaellarabel.com/

                  Comment


                  • #10
                    Originally posted by DanL View Post
                    The 4650 is an older mid-range card, and not "ancient and weak" by any means. Radeon X300 = ancient and weak.
                    You can't buy it in unused condition these days as it's no longer manufactured for a while. It's almost 4 generations old. Hence "ancient".
                    And it's bottom of mid-range or so. When comparing to any modern GPU you can actually buy today, it would be real low end in terms of performance.

                    Comment


                    • #11
                      Any chance that we will one day see a benchmark without using the Unity desktop, which may or may not have an impact on the results?
                      I really doubt it.

                      Comment


                      • #12
                        Originally posted by curaga View Post
                        High-end cpu + low-end gpu is my setup exactly
                        This is very unbalanced setup. It basically shows that you don't really care about 3D performance at all. So, do you really care about 3D benchmarks when running on this setup? Why? I fail to understand that.

                        Comment


                        • #13
                          Originally posted by neatnoise View Post
                          There are no alternative drivers for these cards after amd has dropped support.
                          That's somewhat valid point. Though there is "catalist legacy" which AFAIK suppors older cards and uplifted to work with recent xorg, etc. So it's only partially valid statement. And hey, running other GPUs with opensource drivers is interesting idea as well! After all, I'm really unhappy with quality of proprietary driver and how it behaves. Most of time it would be unable to generate package properly for my system and installing it without package is troublesome and pollutes system to say the least. And deinstall usually wrecked. Should I admit it's a valid reason to dislike proprietary driver? It could wreck ubuntu on nearly every OS version upgrade for example. Especially if you installed version from AMD site. Yet I never faced something like that with opensource drivers.

                          Comment


                          • #14
                            Originally posted by ua=42 View Post
                            Hmm. I bet the reason why he uses a high end CPU when testing graphics cards is so that there won't be weird performance drops because of the cpu being under-powered.
                            Right. However, high CPU usage by driver is a problem as well. And btw, opensource radeon driver is quite troublesome at this. I.e. there is severe CPU load when running 3D games. Attempt to profile this situation lead me to think that it's GPU driver who eats most CPU in it's ioctl. Unfortunately, tracing, profiling and so on has proven to be very delicate when it comes to graphics. I've got a number of hard lockups when trying to profile or trace things anyhow better so I only managed to get impression that ioctl of radeon's driver is a CPU hog. But not more than that.

                            Comment


                            • #15
                              I was investigating why World Of Padman appears to be slower with Mesa 9.1.

                              The fact is it's not slower at all. The problem is the GPU hardlocks twice during the benchmark and the recovery takes 10 seconds, so in total the computer is frozen (0 fps) for 20 seconds. The cause of the hardlocks is Hyper-Z and the reproducibility of the hardlocks is 100%.

                              Comment

                              Working...
                              X