Announcement

Collapse
No announcement yet.

The Extreme Cases Where A Sub-$200 NVIDIA GPU Can Beat A $550+ AMD R9 Fury On Linux

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

  • #11
    I'm sure AMD knows about it. I'm also pretty sure it's not gonna get fixed until the amdgpu work gets done.

    Comment


    • #12
      Grabs popcorn.

      Waits for the Astroturfing to explain everything away.

      Meanwhile, nVidia market share goes thru the roof.... http://www.fudzilla.com/news/graphic...the-gpu-market

      To much spin, not enough reel....

      GreekGeek :-)

      Comment


      • #13
        Michael would you happen to have any xonotic and supertuxkart benchmarks on the that GTX950 for ultra+ settings, by which i mean including extra stuff that just ultra does not apply
        i was about to be blown away then i saw the settings, wondering how it compares to my OCed Galaxy 650 TI Boost with sustained boot clocks of 1154 - 1204Mhz

        Comment


        • #14
          Originally posted by pqwoerituytrueiwoq View Post
          Michael would you happen to have any xonotic and supertuxkart benchmarks on the that GTX950 for ultra+ settings, by which i mean including extra stuff that just ultra does not apply
          i was about to be blown away then i saw the settings, wondering how it compares to my OCed Galaxy 650 TI Boost with sustained boot clocks of 1154 - 1204Mhz
          Not off hand besides what the test profile exposes as that's what was recommended/told to me by the STK developers.
          Michael Larabel
          http://www.michaellarabel.com/

          Comment


          • #15
            Originally posted by gotwig View Post
            I wonder why you even mess with AMD than... Every normal linux gamer has nvidia graphics
            Guess I must not be normal; let me just drop my ethics and get a ton of money so I can become normal

            My 7850 has done well on both Windows and Linux. OpenGL performance could be better if anything, but currently most stuff I've thrown at this card with open-source graphics drivers was fine. I don't play anything really significant though (most graphically intense game I've played maybe was Xonotic; most mainstream native being Dota 2, and most mainstream through Wine being Guild Wars 2).

            As far as I'm concerned though, AMD still has the edge in open-source graphics driver support (and ethics I guess, for whatever that's worth). As long as that remains true, AMD will continue to have my business, but it's real disappointing to see the driver situation with some cards/scenarios.

            Comment


            • #16
              Meanwhile:
              http://arstechnica.com/gaming/2015/0...nt-for-nvidia/

              Time will tell when Vukan API spec is released and implemented.

              Comment


              • #17
                I hope soon AMD will be able to roll out a much-improved Catalyst Linux driver that's able to make it much more competitive for Linux OpenGL workloads -- part of the reason for blatant articles like these is to increase the public pressure for change.
                Oh my, now you even have explination for the blatant articles about AMD you write

                Why don't you apply switch '--eon_disable_catalyst_workarounds' for Bioshock benchmark, someone here on phoronix even mentioned it is fine as of 15.7 driver i think and it should be faster that way. It is a game switch and fglrx users are informed by that... Well maybe for DiRT it might be also needed, but dunno for sure.... well, maybe for all eON wrapped games is fine now to run with that switch, because name of the switch sounds like that to me

                Well still not as fast as nVidia because most eONs also has __GL_THREADED_OPTIMIZATIONS=1 in a script ... which means for Catalyst you might try to rename those binaries to some of the MCCaps1-enabled profile title, for example as 'openarena.x86_64' or 'Borderlands2'... MCCapsX is what enable threaded optimizations in fglrx, well only for selected games where it is proven that it work... and there is not just one in fglrx, but 3 variants of those threaded optimizations

                So nope AMD does not give users env vars which are proven that does not always work fine, but yeah if profile is missing users can already play with those by renaming
                Last edited by dungeon; 20 August 2015, 08:57 PM.

                Comment


                • #18
                  At this point, anyone serious will get a card from Nvidia.

                  I believe it's about 7 years since AMD should offer good drivers, but they seem to struggle more for each new generation. I wonder how they should be able to keep up with Vulkan support in addition.

                  Comment


                  • #19
                    Originally posted by gotwig View Post
                    I wonder why you even mess with AMD than... Every normal linux gamer has nvidia graphics
                    The open-source drivers for AMD cards are way ahead of the ones for NVidia ones, and some of us really don't like binary blobs.
                    Or, in the other direction, DX12: http://arstechnica.com/gaming/2015/0...nt-for-nvidia/

                    Comment


                    • #20
                      Originally posted by Xaero_Vincent View Post
                      I'd like to hear our friendly neighborhood AMD guy's (bridgman) take on this?
                      I'm expecting Michael to write another article tomorrow about the case from (yesterday's ?) benchmark results where the less expensive AMD card was running at the same speed as the more expensive NVidia card.

                      EDIT - that's odd, thought there were more graphs in the follow-up article before ?
                      Last edited by bridgman; 20 August 2015, 09:13 PM.

                      Comment

                      Working...
                      X