Announcement

Collapse
No announcement yet.

NVIDIA Linux Performance-Per-Dollar: What The RX 480 Will Have To Compete Against

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

  • #11
    Originally posted by bridgman View Post
    Hah, that's funny. NVidia and ATI/AMD have both had big R&D centers in Shanghai for a decade or so - we opened in 2006 and they opened in 2007. They were pretty much side-by-side in Zhangjiang High-Tech Industry Park for a while.
    Another article said Polaris/Vega were entirely designed in Shanghai. Aside from maybe an architect or manager in the US, can you refute that?

    Comment


    • #12
      Originally posted by coder View Post
      Crossfire on Linux? Will that ever work?

      Or maybe you just meant for multiple machines.
      If Vulkan becomes widespread, multi-GPU will be in the hands of the game developers. So let's say "something like that" will be supported, but probably won't be used, except for big budget AAA titles with good PC/Vulkan support.

      Comment


      • #13
        Originally posted by coder View Post
        Another article said Polaris/Vega were entirely designed in Shanghai. Aside from maybe an architect or manager in the US, can you refute that?
        Not sure about Polaris (I wasn't really involved with it), but that's definitely not the case for Vega. I don't think it's true for Polaris either.
        Test signature

        Comment


        • #14
          Originally posted by bridgman View Post

          Not sure about Polaris (I wasn't really involved with it), but that's definitely not the case for Vega. I don't think it's true for Polaris either.
          I guess people based that claim on what Raja twitted couple days ago

          Comment


          • #15
            Originally posted by coder View Post
            Crossfire on Linux? Will that ever work?

            Or maybe you just meant for multiple machines.
            i couldn't care less about xfire/sli, only reason why i would go with 2 gpus in 1 machine would be to play with vulkan i plan on looking up. and xfire/sli is not present on vulkan anyway since it is up to application to do it efficiently

            i will most probably replace gpus in all my machines with NVidia just so i can move to wayland there as well. i always only had 1, but recently number jumped to 3. as long as performance is comparable to 759Ti, it is enough to spend 200$ per card

            Comment


            • #16
              Originally posted by eydee View Post

              If Vulkan becomes widespread, multi-GPU will be in the hands of the game developers. So let's say "something like that" will be supported, but probably won't be used, except for big budget AAA titles with good PC/Vulkan support.
              or engines that will facilitate it. it is easy to imagine being implemented in something as UE4 when you consider you can make game without ever writing single line of code.

              Comment


              • #17
                Originally posted by dungeon View Post
                I guess people based that claim on what Raja twitted couple days ago
                I suspect that's the major attraction of Twitter - 140 char isn't enough to say anything unambiguously

                The team pic looks about the right size for "the people in Shanghai who worked on Polaris", but way too small for "all the people who worked on Polaris".
                Test signature

                Comment


                • #18
                  Originally posted by bridgman View Post
                  I suspect that's the major attraction of Twitter - 140 char isn't enough to say anything unambiguously
                  But it seems enough for journalists to add to it There are couple articles around, but all ends up being based on that Raja's twitt

                  Comment


                  • #19
                    Originally posted by bridgman View Post

                    I suspect that's the major attraction of Twitter - 140 char isn't enough to say anything unambiguously

                    The team pic looks about the right size for "the people in Shanghai who worked on Polaris", but way too small for "all the people who worked on Polaris".
                    I'm unsure of why it matters when AMD is an international company...?
                    Is there some conflict with Shanghai that I'm not aware of?

                    Comment


                    • #20
                      Originally posted by bridgman View Post
                      Not sure about Polaris (I wasn't really involved with it), but that's definitely not the case for Vega. I don't think it's true for Polaris either.
                      Sounds like Vega was a whole new architecture instead of a 350mm² RX 480 with HBM...Though it looks quite easy to get on top of the 1080 with small HBM2 connectors in theory...
                      I'm pretty sure that the RX 470 4GiB would win this Perf/Cost efficiency race easily even with current drivers... A 310mm² die for 700$(GTX1080) is more than cheeky in my opinion even in the crippled 1070 version for 450$.
                      I think it's pretty obvious that Nvidia pushed Pascal to an unprepared release without being able to serve. The purpose of the immense price tags is to lower the demand for few weeks(Founders Edition) until enough units can be produced.
                      Originally posted by coder View Post
                      Crossfire on Linux? Will that ever work?
                      No need with Vulkan. Vulkan and Heterogenous computing will solve all our problems, trust me The most important thing we need pretty quickly is Sea/Southern Islands support and 144Hz/FreeSync support in my opinion.
                      The HCCompiler and other OpenGPU Tools should be in the Ubuntu repos to encourage indie devs etc.

                      Comment

                      Working...
                      X