Announcement

Collapse
No announcement yet.

Ubuntu 12.10: Open-Source Radeon vs. AMD Catalyst Performance

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

  • #11
    I don't really care for 3D performance but this is ridiculous. Why is the performance gap that big ? I mean WTF ? They do have paid developers after all.

    Still I do appreciate having a stable open source driver.

    Comment


    • #12
      Originally posted by marek View Post
      I guess the open driver on Ubuntu has neither PCIe 2.0 turned on nor 2D tiling.

      I'm not sure what's needed to get PCIe 2.0. Maybe kernel 3.6? For 2D tiling, you need xf86-video-ati from git, so the benchmark results aren't so surprising.

      That said, I've got no idea how much the results would be different with everything from git.
      I have been doing some tests myself with the latest stuff from Git. Tiling is enabled. PCIe 2.0 is enabled too (you need a kernel parameter radeon.pcie_gen2=1), but it has no effect on my tests, probably because all my textures are static. So, in my tests the latest r600 gallium is about 50% of the fglrx.

      Comment


      • #13
        Pathetic at best.

        Comment


        • #14
          Basically every oss driver is not for "real" gamers. I would not worry much about that if amd would not drop support for older cards for fglrx so soon. Ok, nvidia considered now dx9 cards as legacy but most likely they will update the 304 driver series to newer kernels/xservers still for a while. fglrx already dropped dx10 cards before having at least 1 driver you could use without that many bugs. 12.11 beta was a step in the right direction, maybe not perfect but better than the rest - but dx10 card users will stay with 12-6 beta forever, with all it's bugs - and even for systems with hybrid hd 4200/5400 a new driver does not work.

          Comment


          • #15
            Originally posted by mcirsta View Post
            I don't really care for 3D performance but this is ridiculous. Why is the performance gap that big ? I mean WTF ? They do have paid developers after all.
            Part of the reason is they don't have the manpower to implement everything required to do efficient GPU work, but another part is that they're working with the publicly-available documentation for the GPUs, rather than having access to the hardware implementations that may have hidden optimisations they don't want to broadcast to Nvidia and Intel. The OSS guys have said they're working on at least getting access to the internal GPU emulators that are built as part of the GPU development process, but even with all the community support they can handle (currently they have very little), their target is somewhere between 60-80% (closer to 60%) of the performance of the closed-source driver.

            Comment


            • #16
              So I benchmarked Reaction Quake 3 on r600g. Here are the results:

              With Phoronix Test Suite: 19 fps
              Without Phoronix Test Suite: 121 fps

              I wonder what PTS is doing differently.

              Comment


              • #17
                Originally posted by marek View Post
                So I benchmarked Reaction Quake 3 on r600g. Here are the results:

                With Phoronix Test Suite: 19 fps
                Without Phoronix Test Suite: 121 fps

                I wonder what PTS is doing differently.

                Just in general, I noticed the performance doesn't scale well with OpenGL effects compared to Catalytst. Turning on bloom alone in the new version of openarena is enough to slaughter the framerate by 1/8th on r300g. I don't have the benchmarks with Catalyst and bloom, but I'm pretty sure it doesn't get slaughtered by 1/8th.

                The less OpenGL effects you stack up, the closer the open source driver gets to Catalyst.

                I noticed PTS likes to turn on every graphics option, including non-default graphics options. So pay attention to the graphics option as even a single option in there can sometimes absolutely slaughter the framerate of the open source driver compared to Catalyst.

                It's not that the open source driver is a lot slower, rather it's just a tiny bit slower at some effect and then that effect is getting multiplied by 100x per second which adds up quickly. Turning off the graphics option and the framerate increases multiple fold. For some games, it's just a matter of finding which switch is killing your framerate with the open source driver, and it's almost always an optional graphics option.
                Last edited by Sidicas; 30 October 2012, 11:56 PM.

                Comment


                • #18
                  Originally posted by allquixotic View Post
                  What really galls me about the open source AMD drivers is how much they've stalled on providing proper 3D support for Southern Islands. I thought it was here many months ago, but I was wrong.... it was just a "first cut". AMD doesn't seem to know how to really finish off the support on their open drivers -- they get it to the point where it can render a few triangles, and move on to the next chipset. And I guess the rest of the "community" built up around the open drivers hasn't invested the time this cycle to get SI running.

                  Months ago, I was keeping track of the number of days we went before we could get a true and proper composited 3D desktop experience with working (mostly bug-free and performant) 2D and 3D drivers and OpenGL 2.1. I recall that the first mega-patch for SI hit well in advance of how long it took for Evergreen. Well, now it looks like SI will take longer than Evergreen. Definitely not a good sign, guys.
                  Interestingly enough, a patch set hit the mailing list today which enables control flow on the SI driver. It'd be good to see what the current status of the driver is with that - supposedly i think it's getting pretty close now, but i don't know if it's still crashing all over the place or if it's considered stable.


                  radeonsi can run piglit reliably with it now. It fixes hundreds of tests and prevents many more from crashing.

                  Comment


                  • #19
                    60 fps

                    I wonder what the purpose of 300 fps would be. As far as I have heard the human eye will only register up to 60 fps - so perhaps a better measure would be "at what resolution does the driver performance sink below ideal fps".

                    This might give a different readout than just completely un-needed fps that do not give any noticeable difference for the human user.

                    Comment


                    • #20
                      Originally posted by mcirsta View Post
                      I don't really care for 3D performance but this is ridiculous. Why is the performance gap that big ? I mean WTF ? They do have paid developers after all.

                      Still I do appreciate having a stable open source driver.
                      Well, people have different needs. But I'm glad you realize the realities. This is what we got after five years(?) now of sponsored work. Imagine where we could have been if these developers spent their effort on Fglrx and perhaps AMD focused on open sourcing this instead? The situation for modern OpenGL support is even worse (OpenGL 3.x +), as well as advanced AA.
                      Last edited by efikkan; 31 October 2012, 02:19 AM.

                      Comment

                      Working...
                      X