Announcement

Collapse
No announcement yet.

Radeon Vulkan Driver Benchmarks: AMDVLK 2018.4.2 vs. AMDGPU-PRO 18.40 vs. Mesa 18.2/19.0

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

  • #21
    UPDATE: regression fixed https://patchwork.freedesktop.org/patch/263760/

    That was fast lol

    Comment


    • #22
      Originally posted by BNieuwenhuizen View Post
      We copy some ideas but almost 0 code. There is one shared libary mostly about texture layout rules.
      Ah sorry, my mistake.

      Originally posted by M@yeulC View Post
      The solution would be to make the two codebases gradually closer, and ending up with exactly one codebase.

      But the way AMDVLK is developed with regular code drops isn't really open source friendly, as far as development goes.
      As much as I agree with the solution you propose, it doesn't seem like that's going to be a reality any time soon, and your 2nd sentence is a decent example why. Although I I don't think RADV and AMDVLK will never end in 1 codebase, both drivers seem to have their own sets of pros and cons, where as long as those differences remain, they're going to remain separate. This fragmentation I feel is doing more harm than good (vs having 1 codebase that all current devs revolve around), especially considering that I don't think any of the current devs from either team wouldn't get along. I think the Mesa devs might be a little more stern about the regular code drops, but I'm sure the AMDVLK team could easily adapt to such a thing without too much of a fuss.

      Originally posted by IreMinMon View Post
      Afaik, both AMDVLK and RADV are opensource, so they could at any point copy the code over to their driver. This is competition + collaboration. And is imho very good for end user.
      Yes, hence my 4th point.

      Comment


      • #23
        Originally posted by GruenSein View Post
        Also, they can never relinquish control about releases or risk their code being rejected because their hardware enablement needs to be available when new hardware is released. So, going all Mesa simply isn't an option.
        That's not true because they're already going all mesa for their OpenGL FOSS driver.
        ## VGA ##
        AMD: X1950XTX, HD3870, HD5870
        Intel: GMA45, HD3000 (Core i5 2500K)

        Comment


        • #24
          Originally posted by tichun
          Stop spreading lies that it isnt open source or that one cannot contribute
          Just use README.md to understand how to contribute.
          Dude calm down and reread my post; It's just a question. You should learn how to read.

          Comment


          • #25
            Originally posted by smitty3268 View Post

            That seems rather unnecessarily insulting, given that there are IMO some pretty reasonable concerns about the contribution process and distribution of amdvlk. And it's obviously tough to just abandon something you've worked hard on until you can really show it to be inferior - imagine throwing away radeonsi at this point in favor of a newly open sourced fglrx, for example.

            Not to mention the fact that to date, radv has provided faster hardware support for every new card AMD has released than amdvlk. I realize bridgman keeps saying that next time it will be better, but we've been hearing that for a long time now and it hasn't happened yet...
            Exactly, I'm pretty disappointed by the way AMD has handled Vulkan driver development so far.
            Contributing to RADV should be a no brainer considering the success of RadeonSI.
            And don't bring up cost: RADV is already a very mature driver and it currently costs AMD literally *nothing*.

            I'm really curious about Intel entering the discrete GPU market with open-source drivers and adaptive-sync support.
            Should bring interesting competition to AMD.

            Comment


            • #26
              Originally posted by smitty3268 View Post
              Not to mention the fact that to date, radv has provided faster hardware support for every new card AMD has released than amdvlk. I realize bridgman keeps saying that next time it will be better, but we've been hearing that for a long time now and it hasn't happened yet...
              Actually I don't think I have *ever* said that, let alone "kept saying" it.

              What I have been talking about is degree of open source driver support in general at launch, and most of the time I was saying it was years ago.
              Test signature

              Comment


              • #27
                Originally posted by bridgman View Post

                Actually I don't think I have *ever* said that, let alone "kept saying" it.

                What I have been talking about is degree of open source driver support in general at launch, and most of the time I was saying it was years ago.
                There was literally a presentation by someone from AMD - maybe Alex? - comparing radv and amdvlk, and one of the bullet points under the amdvlk pros list was faster hardware support to which i could only lol. Sorry if I conflated that with your more general posts about launch time support.

                Anyway, my point was very much that I feel like a lot of things have been said about how good amdvlk will be, but it's an open question how much that can be trusted. I'm not saying this as an attack by any means, so please don't take it that way. But when you have things like the recent 590 card launch that fails to run on linux, and before that the major problems with Raven support, and the still to be worked on nature of GCN 1.0 vulkan/amdgpu support I think it's a very tough sell for AMD to say "trust us everything will be fine going forward and there's no need for anyone else to be able to do this themselves instead".

                To be fair, all these issues are obviously much better than what was going on 5 years ago. I expect they'll continue to get better. It just doesn't feel like the driver situation on linux for AMD has really fully arrived at a place where it can always be given the benefit of the doubt yet. For me at least, it's in a "i'll believe it when i see it" state. As I mentioned in a previous post, I'm starting to see it in regards to performance so that's promising.
                Last edited by smitty3268; 26 November 2018, 04:33 PM.

                Comment


                • #28
                  Originally posted by smitty3268 View Post
                  There was literally a presentation by someone from AMD - maybe Alex? - comparing radv and amdvlk, and one of the bullet points under the amdvlk pros list was faster hardware support to which i could only lol. Sorry if I conflated that with your more general posts about launch time support.
                  Ahh, you're probably talking about Nicolai's presentation at FOSDEM. As far as I know that has been pretty much the only public comments from AMD.

                  We haven't launched a new HW generation since both radv and AMDVLK have been out so it's hard to say what will happen. If things go well we will make the question academic by upstreaming radeonsi support sufficiently early and sampling HW sufficiently well to NDA customers with developers working on radv that both will be ready in time for launch anyways.
                  Last edited by bridgman; 26 November 2018, 07:12 PM.
                  Test signature

                  Comment


                  • #29
                    Originally posted by marek View Post

                    The initial intentions were good and there was void at the beginning that needed to be filled. Now it's just about egos and tribalism.
                    Triballi maybe, Id and Ego too, but Super-Ego nowhere... long time forgotten

                    When in China RX 570 could be 580 so why not, everybody would have clear picture on steam survey anyway

                    Last edited by dungeon; 26 November 2018, 08:34 PM.

                    Comment


                    • #30
                      Originally posted by darkbasic View Post

                      That's not true because they're already going all mesa for their OpenGL FOSS driver.
                      Are they now? I beg to differ. There still is the OpenGL-Pro driver even though Mesa now has compatibility profiles and such.

                      Comment

                      Working...
                      X