Announcement

Collapse
No announcement yet.

More Radeon RX 590 Ubuntu Benchmarks - See How Your Linux GPU Performance Compares

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

  • More Radeon RX 590 Ubuntu Benchmarks - See How Your Linux GPU Performance Compares

    Phoronix: More Radeon RX 590 Ubuntu Benchmarks - See How Your Linux GPU Performance Compares

    Published on Friday was my Radeon RX 590 Linux benchmarks now that the kinks in the support for this latest Polaris refresh are worked out (at least in patch form). Here are some complementary data points with some of the OpenGL tests outside of the Steam games for those curious about the RX 590 performance in other workloads or wanting to see how your own GPU performance would compare to these results...

    http://www.phoronix.com/scan.php?pag...590-More-Linux

  • #2
    R9 390x: 1090MHz GPU - 1500MHz GDDR5 - TDP limit: 225w

    Polaris is weird and pushing clocks will not improve the situation on that node. More cores and lower clocks would be optimal.

    Comment


    • #3
      Originally posted by prazola View Post
      Polaris is weird and pushing clocks will not improve the situation on that node. More cores and lower clocks would be optimal.
      They are more expensive to manufacture, though.

      Comment


      • #4
        Originally posted by tildearrow View Post

        They are more expensive to manufacture, though.
        Well, a partial shrink of a clearly limited architecture is a waste of money too. And looking at power consumption the card performs only as a better binned one.
        [email protected] would probably sell more than the RX590. Whou would ever consider this card today (at this price)?

        Comment


        • #5
          Originally posted by prazola View Post
          More cores and lower clocks would be optimal.
          Which is what Fury and Vega is (sans HBM and some other details), but at the same time, if you compare 56 vs 64 CU's either Fury or Vega at same frequency and power budget then you can see there's little difference in performance from the last extra CU's which means there are diminishing returns from extra cores as well.

          Comment


          • #6
            Originally posted by prazola View Post
            Well, a partial shrink of a clearly limited architecture is a waste of money too.
            The thing with 14 and 12nm is that there's no difference in density. The only thing they did was to shrink the transistors a bit which increases the empty space between them, giving a little extra thermal and frequency headroom. It's not really the same thing as a full generational leap like 14 ->7nm, it's just a slight optimization of the same node.

            Comment


            • #7
              Originally posted by Brisse View Post

              The thing with 14 and 12nm is that there's no difference in density. The only thing they did was to shrink the transistors a bit which increases the empty space between them, giving a little extra thermal and frequency headroom. It's not really the same thing as a full generational leap like 14 ->7nm, it's just a slight optimization of the same node.
              Yep, that's why they should sell the card as rx 580+. With the actual name it really seems a scam. I really hope AMD is doing something good in its labs because I want to upgrade to another AMD card but there's nothing I consider good enough right now.

              Comment


              • #8
                Originally posted by prazola View Post
                Yep, that's why they should sell the card as rx 580+. With the actual name it really seems a scam. I really hope AMD is doing something good in its labs because I want to upgrade to another AMD card but there's nothing I consider good enough right now.
                Naming schemes from most hardware manufacturers has long been shady as shit. It was a similar situation when they rebranded RX4xx -> RX5xx and the technical differences that time were even more trivial than this time while the branding made it look like a generational leap.

                Also, regarding AMD and branding, there's this recent incident: https://youtu.be/A5Xm2V8O5po

                Comment


                • #9
                  Originally posted by Brisse View Post

                  Also, regarding AMD and branding, there's this recent incident: https://youtu.be/A5Xm2V8O5po
                  A similar thing also happened in 2016 with the Radeon RX 470D, which is a little less than a 470 and was China-only.

                  Comment


                  • #10
                    Originally posted by prazola View Post
                    Yep, that's why they should sell the card as rx 580+. With the actual name it really seems a scam. I really hope AMD is doing something good in its labs because I want to upgrade to another AMD card but there's nothing I consider good enough right now.
                    LOL. Do you think the RX570 should be named the 580-?

                    Bumping the # by 10 for a clock speed increase is pretty much standard practice. It's why the 560, 570, and 580 are different numbers in the first place. It's not like they called this card a 680.

                    Comment

                    Working...
                    X