Announcement

Collapse
No announcement yet.

On Low-End GPUs, Nouveau Speeds Past The NVIDIA Driver

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

  • On Low-End GPUs, Nouveau Speeds Past The NVIDIA Driver

    Phoronix: On Low-End GPUs, Nouveau Speeds Past The NVIDIA Driver

    While the Linux 2.6.38 kernel has been out for less than one week, if you use NVIDIA graphics, particularly with a low-end GPU, start counting down the days to the release of the Linux 2.6.39 kernel. Particularly on lower-end NVIDIA GeForce graphics cards, the reverse-engineered open-source Nouveau driver now meets or exceeds the speed of NVIDIA's official proprietary driver in a number of OpenGL test cases...

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

  • #2
    Originally posted by phoronix View Post
    Phoronix: On Low-End GPUs, Nouveau Speeds Past The NVIDIA Driver

    While the Linux 2.6.38 kernel has been out for less than one week, if you use NVIDIA graphics, particularly with a low-end GPU, start counting down the days to the release of the Linux 2.6.39 kernel. Particularly on lower-end NVIDIA GeForce graphics cards, the reverse-engineered open-source Nouveau driver now meets or exceeds the speed of NVIDIA's official proprietary driver in a number of OpenGL test cases...

    http://www.phoronix.com/vr.php?view=OTIzMg
    Holy crap!!
    I am becoming to think that nvidia GPUs, are somewhat easier to command(program) than Radeons are. I have no other explanation for this performance boost without any nvidia support, or documentation. Maybe some one of the Radeon developers will clear the situation.

    Comment


    • #3
      Is there any change you could postfix any prerelease kernels with -pre or -rc as it get's confusing when you're talking about testing 2.6.39 when rc1 hasn't even been tagged yet

      Look at the number of articles you have testing 2.6.38 which has only been out for a short time

      Comment


      • #4
        Go Nouveau!

        It's great that even Nvidia users have a choice!

        Comment


        • #5
          Could it be that you are using a high-end CPU in the testing system?
          Testing a Geforce 8500 on a six-core i7 with a gpu driver that is mostly CPU bound seems pointless to me ...

          Comment


          • #6
            Originally posted by Flyser View Post
            Could it be that you are using a high-end CPU in the testing system?
            Testing a Geforce 8500 on a six-core i7 with a gpu driver that is mostly CPU bound seems pointless to me ...
            That's exactly what wouldn't happen. A low-end GPU on top of a high-end CPU would mostly likely - not - stop at the driver's CPU bottlenecks.

            Comment


            • #7
              Originally posted by Daekdroom View Post
              That's exactly what wouldn't happen. A low-end GPU on top of a high-end CPU would mostly likely - not - stop at the driver's CPU bottlenecks.
              Do you have any explanation of proof for that?

              Comment


              • #8
                Originally posted by Flyser View Post
                Do you have any explanation of proof for that?
                If this test was CPU bound, all results including 9800GT would be pretty much the same. If the 9800GT managed to get ahead, it's because the CPU wasn't a bottleneck for any card that underperformed it.

                Comment


                • #9
                  Don't forget the nouveau rendering lacks anisotropic and anti aliasing. Unfortunately PTS uses the high quality nexuiz preset which is not fully supported by open drivers, so comparisons are useless.
                  ## VGA ##
                  AMD: X1950XTX, HD3870, HD5870
                  Intel: GMA45, HD3000 (Core i5 2500K)

                  Comment


                  • #10
                    I may be missing something, but is this all about the open source driver scoring 20.52fps versus nvidia binary scoring "only" 19.95fps? In one test?

                    Comment


                    • #11
                      Originally posted by Drago View Post
                      Holy crap!!
                      I am becoming to think that nvidia GPUs, are somewhat easier to command(program) than Radeons are. I have no other explanation for this performance boost without any nvidia support, or documentation. Maybe some one of the Radeon developers will clear the situation.

                      Actually they are!

                      AMDs architecture takes packets of 5 ( or 4 on Cayman) Commands per Streaming Processor.
                      The Nvidia equivalent takes only one command.

                      So the AMD driver needs to find 5 (4) Commands to be packaged together.
                      The Nvidia driver just needs to give them fire.

                      Comment


                      • #12
                        i'm curious to see more results, such as seeing one of the low-end fermis vs high end, and i'm more interested in the results of the geforce 7000 series since that was the last dx9 series.

                        Comment


                        • #13
                          Originally posted by bug77 View Post
                          I may be missing something, but is this all about the open source driver scoring 20.52fps versus nvidia binary scoring "only" 19.95fps? In one test?
                          yes you are, that "the reverse-engineered open-source Nouveau driver now meets or exceeds the speed of NVIDIA's official proprietary driver in a number of OpenGL test cases..." means it was not as fast as the nvidia binary scoring before, not that it now beats it at a few .% now...

                          they may even optimise today's code some more by Monday given encouragement from this article OC, but i dont suppose he will pull a new git and run tests by then if they do

                          Comment


                          • #14
                            The nouveau driver would definitely be ideal for those cards that aren't supported by the blob any longer, such as the GeForce 5 and older. The GeForce 6 and 7 series can also benefit too. At some point it'll be possible to support the newest cards once the devs figure out how to turn on the antialiasing modes and to make use of the stream processors. I have to admit that the Nouveau driver has made great strides despite not having any support or documentation from nVidia.

                            Comment


                            • #15
                              Once again, no CPU usage graphs which makes the whole comparison kinda useless.

                              Michael, you promised them, remember?

                              Comment

                              Working...
                              X