Announcement

Collapse
No announcement yet.

The Performance-Per-Watt, Efficiency Of GPUs On Open-Source Drivers

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

  • The Performance-Per-Watt, Efficiency Of GPUs On Open-Source Drivers

    Phoronix: The Performance-Per-Watt, Efficiency Of GPUs On Open-Source Drivers

    To complement the initial results earlier this week of trying 60+ graphics cards on the open-source Linux GPU drivers, here's the second round of our mass open-source graphics driver testing. While in Wednesday's article were the raw OpenGL results for the wide-range of graphics processors on the open-source Intel, Radeon, and Nouveau articles, in today's article are complementary results providing a brief look at the system power consumption, performance-per-Watt, CPU usage, and GPU thermal information when testing the hardware in the same configuration.

    http://www.phoronix.com/vr.php?view=20511

  • #2
    Michale, why do you even bother to include the Radeon R7 260X if you don't want to update the firmware? This is a trivial fix.

    Comment


    • #3
      Was radeon.dpm enabled for the modern ATI/AMD cards?

      Comment


      • #4
        Originally posted by MannerMan View Post
        Was radeon.dpm enabled for the modern ATI/AMD cards?
        dpm won't make much difference when your objective is to test the "full out" performance, since it will automatically scale up to full power.

        Comment


        • #5
          There is a major flaw in this test that was nearly alluded to in the write-up; when the gpu performance goes up, the *whole system* power consumption goes up in order to be able to feed the GPU as much data as it is able to consume. It would be very helpful to be able to test the power consumed *specifically* by the GPU -- which may be possible on discrete units that have their own power plug.

          It is pretty impressive that the best performing GPUs are also able to generate the lowest total system power consumption... though I suspect that if it was testing GPU power consumption specifically, the difference would be even greater.


          Anyhow... AMD FTW!

          Comment


          • #6
            Originally posted by droidhacker View Post
            dpm won't make much difference when your objective is to test the "full out" performance, since it will automatically scale up to full power.
            True that. Guess idle power-usage would be more interesting with/without DPM.

            Comment


            • #7
              Or the cards being on the "battery" profile. Right now the comparison between radeon and nouveau doesn't really make sense, as the former is pretty much on performance, and the latter on powersave...

              The heat graph was pretty interesting, though. I still don't get how the 4890 has lower heat than the 5450...

              Comment


              • #8
                Originally posted by GreatEmerald View Post
                The heat graph was pretty interesting, though. I still don't get how the 4890 has lower heat than the 5450...

                The HD 5450 was passively cooled while the ASUS HD 4890 has a very large cooler.
                Michael Larabel
                http://www.michaellarabel.com/

                Comment


                • #9
                  Originally posted by droidhacker View Post
                  dpm won't make much difference when your objective is to test the "full out" performance, since it will automatically scale up to full power.
                  Besides, the AMD cards still came out on top in terms of PPW. Speaking of which, I'm actually pretty surprised intel's GPUs are relatively inefficient.

                  Comment


                  • #10
                    Interesting benchmark.

                    How about a comparison with the proprietary drivers (probably just to cry a little)?

                    Comment


                    • #11
                      Originally posted by Calinou View Post
                      Interesting benchmark.

                      How about a comparison with the proprietary drivers (probably just to cry a little)?
                      As said already, the proprietary tests are coming in later articles.
                      Michael Larabel
                      http://www.michaellarabel.com/

                      Comment


                      • #12
                        Originally posted by droidhacker View Post
                        dpm won't make much difference when your objective is to test the "full out" performance, since it will automatically scale up to full power.
                        I wouldn't be so sure about this. DPM also uses power gating or clock gating to power down blocks that might not be used by typical graphics loads, or which are not fully loaded all the time. In addition, DPM also is used for adaptive clocking (powertune and turbo modes). Maybe some of the AMD devs can comment.

                        Comment


                        • #13
                          Originally posted by brent View Post
                          I wouldn't be so sure about this. DPM also uses power gating or clock gating to power down blocks that might not be used by typical graphics loads, or which are not fully loaded all the time. In addition, DPM also is used for adaptive clocking (powertune and turbo modes). Maybe some of the AMD devs can comment.
                          Also "as fast as the driver can feed the hardware with commands, geometry and textures" isn't necessarily the same as "as fast as the hardware can run", although for 260X there probably isn't a huge difference.

                          Comment


                          • #14
                            Originally posted by MannerMan View Post
                            True that. Guess idle power-usage would be more interesting with/without DPM.
                            i think that would be why CPU workload was included. however... i MIGHT be interesting to have everything in one graph, like with columns of different thickness overlaying or whatever having different colours like

                            ____________________
                            ============-----|
                            =======--------------|
                            ___________________ |

                            _______________
                            ========-----|
                            ====------------|
                            ______________ |




                            so one can see many results to compare and to see total and relative stuff at once (like fps - thick bar - and CPU usage and power consumption - smaller bars) - while consumton per frame would be the most important.
                            it would very much show overhead too. so one could see the relation of these too. throughout the drivers GPUs and maybe different systems too.

                            i wanted to to have then overlayed with colours anyway... just a possibility.
                            Last edited by jakubo; 06-05-2014, 02:09 PM.

                            Comment


                            • #15
                              wth... that not what i was referring to..?!
                              i meant that the GPU power consumption would be vaguely described by the total consumption - cpu consumption (workload)

                              Comment

                              Working...
                              X