Announcement

Collapse
No announcement yet.

How good can the DRI drivers be?

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

  • #11
    upgrade? well, build mesa, new xorg-server (if an update is available) build x11-drm modules, and then build the driver.

    since we're on the topic of the open drivers - this site really needs some attention for people interested in opensource ati drivers development :



    maybe there will be some info on r500 sometime in the future. (jerome glisse recently gave a small r500 status update on dri-devel mailing list).

    Comment


    • #12
      Originally posted by Michael View Post
      The DRI driver performs less than the fglrx driver. See: http://www.phoronix.com/vr.php?view=8656
      Heh... That's not apples to apples. You're comparing the DRI driver R300-to-R300 and AMD's got a little of an edge on the DRI team there- beings that AMD understands the chip better. They don't understand Linux as well, unfortunately, so the driver is sub-par compared to Windows, which is, sadly, somewhat sub-par compared to NVidia. Now, the right way to do these comparisons is to do R200 to whatever... For SOME applications, the R200 support, because much more robust T&L support than the "comparable" R300 chips' support provided by the closed source drivers , will do better (MUCH BETTER) with an R200 with DRI than the closed source R300 answer.

      Michael, I'd LOVE for you to do a review of the thing I just mentioned- or to let me do a guest write-up on the subject.

      Comment


      • #13
        Originally posted by Svartalf View Post
        Heh... That's not apples to apples. You're comparing the DRI driver R300-to-R300 and AMD's got a little of an edge on the DRI team there- beings that AMD understands the chip better. They don't understand Linux as well, unfortunately, so the driver is sub-par compared to Windows, which is, sadly, somewhat sub-par compared to NVidia. Now, the right way to do these comparisons is to do R200 to whatever... For SOME applications, the R200 support, because much more robust T&L support than the "comparable" R300 chips' support provided by the closed source drivers , will do better (MUCH BETTER) with an R200 with DRI than the closed source R300 answer.

        Michael, I'd LOVE for you to do a review of the thing I just mentioned- or to let me do a guest write-up on the subject.
        We have done the R200 benchmarking in the past and the R200 could be added to the TODO list, which is very long as you may imagine.

        We do also accept freelance articles at Phoronix.
        Michael Larabel
        https://www.michaellarabel.com/

        Comment


        • #14
          Originally posted by Michael View Post
          We have done the R200 benchmarking in the past and the R200 could be added to the TODO list, which is very long as you may imagine.

          We do also accept freelance articles at Phoronix.
          I'm going to take that as a hint, then...

          (I've got a bit of a todo list as well, so it might be a couple of weeks yet...)

          Comment


          • #15
            Yes, things have changed since then, and now the r200 DRI driver is on par, performance-wise, with the last releases of fglrx that had working r200 support (its image quality should be a bit better as well)

            And yes, r200 DRI support is way better than r300 DRI support, comparatively (notice I'm not saying r200s are faster than r300s here, just that there's more left to optimize in the r300 dri driver).

            If you want to get the maximal performance with the DRI drivers, you have to tweak things a bit, in particular I suggest you enable hyperz. There is a tool called driconf (which your distro probably packages) that lets you do that in a graphical fashion.

            Comment

            Working...
            X