Announcement

Collapse
No announcement yet.

Deus Ex: Mankind Divided On Linux With Latest RadeonSI - Up To 2~3x Faster

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

  • Deus Ex: Mankind Divided On Linux With Latest RadeonSI - Up To 2~3x Faster

    Phoronix: Deus Ex: Mankind Divided On Linux With Latest RadeonSI - Up To 2~3x Faster

    With Marek's latest set of RadeonSI Gallium3D patches, which are said to improve the Deus Ex: Mankind Divided performance by around 70%, having landed in Mesa Git, here are some fresh benchmarks with a Radeon RX 480 and R9 Fury...

    http://www.phoronix.com/scan.php?pag...nuary-2017-RSI

  • #2
    wow! thats awesome to see. Thanks for this

    ----
    btw - any rocket league changes with this?

    Comment


    • #3
      So 70% or Up To 2~3x Faster for RX 480 actually means... 33% improvment on low and 11% degradation on ultra

      Yeah Mesa seems somewhat prefer big GPUs like that Fury, but Mario's princess is in another castle since majority and that every statistic show about 85% of people does not buy big GPUs, so if mesa is not for majority why it is default in distros other than for being opensource?

      So with these 70%, majority will have this expresion






      But OK 30+ fps is playable i guess, again that is minimal/recommended target even per Feral's system requirements
      Last edited by dungeon; 01-06-2017, 09:12 PM.

      Comment


      • #4
        Originally posted by dungeon View Post
        So 70% or Up To 2~3x Faster for RX 480 actually means... 33% improvment on low and 11% degradation on ultra

        Yeah Mesa seems somewhat prefer big GPUs like that Fury, but Mario's princess is in another castle since majority and that every statistic show about 85% of people does not buy big GPUs, so if mesa is not for majority why it is default in distros?

        So with these 70%, majority will have this expresion



        But OK 30+ fps is playable i guess, again
        You are apparently making that statement off of this ones instance... in the past Mesa has been more efficient on lower end GPUs as developers have not had as many high end GPUs to test on... and inefficiencies in the CPU side of things tends to mean that the driver can get CPU bound/blocked such that a high end card cannot reach it's full potential but a lower end one can be driven just fine albeit with higher CPU usage than the proprietary driver.

        Comment


        • #5
          Originally posted by cb88 View Post
          You are apparently making that statement off of this ones instance...
          I am not making anything, even article shows as for one card it goes like this and for another it goes different, even in opposite direction.

          Also other articles shows same tendency with Fury.
          Last edited by dungeon; 01-06-2017, 09:09 PM.

          Comment


          • #6
            Originally posted by dungeon View Post

            I am not making anything, even article shows as for one card it goes like this and for another it goes different, even in opposite direction.
            "Yeah Mesa seems somewhat prefer big GPUs like that Fury" .... that has never been the case. Even now the only reason the RX480 is doing better is because of a regression.... and that the RX480 is literally a much slower card no matter the driver.

            The Fury is still well below where it should be performance wise relative to proprietary drivers... it should be up around the Geforce 980...

            Comment


            • #7
              Originally posted by cb88 View Post
              "Yeah Mesa seems somewhat prefer big GPUs like that Fury" .... that has never been the case. Even now the only reason the RX480 is doing better is because of a regression.... and that the RX480 is literally a much slower card no matter the driver.
              If driver does not matter, then why drivers makes any difference?

              The Fury is still well below where it should be performance wise relative to proprietary drivers... it should be up around the Geforce 980...
              ... and lets not forget that this game should be 30% faster even on NVidia like on Windows No one really target nvidia driver here, as at least on paper driver might happen to be even faster in some cases, basically nvidia driver perf is neither the limit nor target

              I am not commenting where things should be (to be honest i don't expect scalibilty among vendors and different drivers as that is really rare) but what it is and according to what article shown to us
              Last edited by dungeon; 01-06-2017, 09:48 PM.

              Comment


              • #8
                Weird, I might have to profile Deus Ex on Ultra settings and see what shows up. I'll probably use 285, because I don't have 480.

                That Fury performance is just pure luck.

                For Rocket League, see the bug report.

                Comment


                • #9
                  Originally posted by dungeon View Post

                  If driver does not matter, then why drivers makes any difference?



                  ... and lets not forget that this game should be 30% faster even on NVidia like on Windows No one really target nvidia driver here, as at least on paper driver might happen to be even faster in some cases, basically nvidia driver perf is neither the limit nor target

                  I am not commenting where things should be (to be honest i don't expect scalibilty among vendors and different drivers as that is really rare) but what it is and according to what article shown to us
                  That was a typo... I mean The only reason the RX480 isn't doing better...

                  But whatever, I have an RX480 ... yes it is a slower card than a Fury but a lot cheaper also. I just think it is stupid for people to complain about this driver as good as it is. Certainly there aren't any developers ignoring lower end cards .... as far as I can tell they have gotten alot of support from the Mesa developers that never would have been there in the propietary drivers.

                  Comment


                  • #10
                    I should point out a typo;
                    The RX 480 was 1.5x faster while the R9 Fury was 3.5x faster!
                    The R9 Fury was actually 2.4x as fast in the new test.

                    Comment

                    Working...
                    X