Announcement

Collapse
No announcement yet.

AMD FOSS developers working (almost) full time on Vulkan?

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

  • AMD FOSS developers working (almost) full time on Vulkan?

    I'm just speculating here...

    It seems like the speed of especially the AMD FOSS drivers have slowed down quite a lot during the last half year or so, do you think it's possible that they are spending most of their time working on a Vulkan implementation based on the Mantle design (which must be something they have had access to) so they are able to release it at the same time it is made public? Since a lot of the Vulkan design will be based on Mantle, I assume that first implementing a Mantle-esque design and then adapt it to the Vulkan specification is a lot quicker than starting from scratch the day the Vulkan API is locked down/released. Well, of course the devs are a part of the Vulkan design (for instance Graham Sellers is one of the main developers) work and they have access to it during its design phase, but implementing the parts of Mantle that Khronos will keep is probably a quick way of having parts done before it is actually covered by the specification.

    Or is the slower development because they are working on the new AMD driver design? Or am I just delusional, and they are working as quickly as ever on the Radeon/RadeonSI drivers? It just feels like there was nothing really new in Linux 3.18 or 3.19 for Radeon cards (I'm running with an R290) and not much happened on the Mesa side either.

  • #2
    Doubt it, maybe wishful thinking. I have an R9 290 too.

    Also according to Valve developing Vulkan drivers is going to be easier than developing openGL drivers has been.

    Comment


    • #3
      Originally posted by humbug View Post
      Doubt it, maybe wishful thinking. I have an R9 290 too.
      Also according to Valve developing Vulkan drivers is going to be easier than developing openGL drivers has been.
      Hehe, yes, definitely some wishful thinking. But it would be pretty logical, to keep with the times and not have to play catchup on at least ONE graphical API. Especially since the development will be a lot easier, like you said. That is one of the real purposes with Mantle, together with more stringent design and less (driver) overhead.

      Comment


      • #4
        You don't have to guess. git clone mesa, kernel, run some queries, pipe with sed/etc and graph team red commits over time in libreoffice.

        Comment


        • #5
          Originally posted by curaga View Post
          You don't have to guess. git clone mesa, kernel, run some queries, pipe with sed/etc and graph team red commits over time in libreoffice.
          Interesting point, I might just have to do that...

          Comment

          Working...
          X