Announcement

Collapse
No announcement yet.

A Closer Look At The AMDVLK vs. RADV vs. AMDGPU-PRO Vulkan Performance

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

  • A Closer Look At The AMDVLK vs. RADV vs. AMDGPU-PRO Vulkan Performance

    Phoronix: A Closer Look At The AMDVLK vs. RADV vs. AMDGPU-PRO Vulkan Performance

    Back on Christmas Eve I posted our initial AMDVLK Radeon Linux driver benchmarks for this newly open-sourced official Radeon Vulkan driver. Complementing those earlier Vulkan Linux gaming numbers are some more performance metrics for AMDVLK compared to the Mesa-based RADV driver and then the closed-source AMDGPU-PRO Vulkan driver.

    http://www.phoronix.com/vr.php?view=25733

  • #2
    Why only test only the 2 most recent generation of cards? Both HD 7000 and HD 8000 support Vulkan. Could you throw one of these cards in the mix next time?

    Comment


    • #3
      slacka I don't have any "HD 8000" series hardware but anyhow only have so much time in a day to run tests....
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #4
        Ah, it's a little sad AMDVLK came out rather late. I know, internal IP review and all that stuff, but it would probably have been even more beneficial at an earlier stage. Now we have two drivers which are quite the same in terms of results (plus the blob), or with light differences. I still hope for some kind of merge, no matter how easy people say they can be switched on a running system. In the end that might be picking the raisins / benefits from both and saving some developing power (that could be used elsewhere or even on the Vulkan side itself). Redundancy can be good, but maybe not always.
        Stop TCPA, stupid software patents and corrupt politicians!

        Comment


        • #5
          Originally posted by Adarion View Post
          Ah, it's a little sad AMDVLK came out rather late. I know, internal IP review and all that stuff, but it would probably have been even more beneficial at an earlier stage. Now we have two drivers which are quite the same in terms of results (plus the blob), or with light differences. I still hope for some kind of merge, no matter how easy people say they can be switched on a running system. In the end that might be picking the raisins / benefits from both and saving some developing power (that could be used elsewhere or even on the Vulkan side itself). Redundancy can be good, but maybe not always.
          If AMDVLK is supposed to stay relatively in sync with the windows version, then RADV will probably win the race.

          Comment


          • #6
            RADV at same level or faster good job amd once again

            Comment


            • #7
              This forum is so good about bitching in any situation. Supporting the driver would mean better collaboration with the manufacturer-funded developers. Even if the driver is slower in certain scenarios, there's no reason it can't be improved and made mainstream. RADV may not disappear immediately, if at all, since competitive drivers are actually a good thing in this scenario as it drives better quality. Whichever one you decide to use is up to you, it's not a big deal until one starts getting more developer support than the other. And developers are going to support whatever they want regardless of what users think.

              Comment


              • #8
                Some interesting variations going on with the Vega card. Seems that's where most the optimisations went, hopefully they pass onto NAVI when that comes out because I doubt I'll ever see a Vega card given no ITX options exist and allot of them run very high TDP, which would require me to down-clock the vcore for my ITX Ryzen build (not to mention 350W+ at peak produces ALLOT of heat).

                Comment


                • #9
                  Originally posted by computerquip View Post
                  This forum is so good about bitching in any situation. Supporting the driver would mean better collaboration with the manufacturer-funded developers. Even if the driver is slower in certain scenarios, there's no reason it can't be improved and made mainstream. RADV may not disappear immediately, if at all, since competitive drivers are actually a good thing in this scenario as it drives better quality. Whichever one you decide to use is up to you, it's not a big deal until one starts getting more developer support than the other. And developers are going to support whatever they want regardless of what users think.
                  I think the biggest problem comes from games dev. Which driver should they test against? Will they willingly support both and submit patches to both codebase? That seems unlikely...

                  Comment


                  • #10
                    Originally posted by geearf View Post

                    I think the biggest problem comes from games dev. Which driver should they test against? Will they willingly support both and submit patches to both codebase? That seems unlikely...
                    Agreed. Competition is good, but not with drivers. The community should pick one and cannibalize the other and AMD should help along.

                    Comment

                    Working...
                    X