9 Replies Latest reply on Jul 3, 2015 7:38 PM by black_zion

    Catalyst Power Limit settings

    kingfish

      I did some testing using Furmark with different power limit settings. AMD R9 290. First with power limit set to +50 which disables the throttling feature and then at the default (0) energy saving setting. The Control panel 3D settings are my normal game settings..only the power limit changed.

       

      With default (0) setting:                          With +50 (max) setting

      Min fps = 58                                       Min fps = 69        = +15.9%

      Avg fps = 64                                      Avg fps = 70        = +8%

      Score = 3845                                    Score = 4214      = +9%

       

      Benchmark comparisons > FurMark Scores, Comparative Charts (v1.9, 1.10, 1.13, 1.14) | Geeks3D 

      (notice the crossfire benchmarks with and without the max power limit setting) ..Example>

       

      Sapphire Radeon HD 6970 CF (GPU@880MHz, mem@1375MHz, max GPU temp:78°C, two GPUs / CrossFire ON), TPC: 87W(PSU1) + 33W(PSU2) /430W(PSU1) + 240W(PSU2)

      (on H67 testbed v2), Catalyst 12.3 whql (Win 7 64-bit)
      4704 points (78 FPS), PowerTune: +20%

       

      Sapphire Radeon HD 6970 CF (GPU@880MHz, mem@1375MHz, max GPU temp:78°C, two GPUs / CrossFire ON), TPC: 87W(PSU1) + 33W(PSU2) /352W(PSU1) + 176W(PSU2)

      (on H67 testbed v2), Catalyst 12.3 whql (Win 7 64-bit)
      3505 points (58 FPS), PowerTune: 0%

       

      [Tested] Radeon HD 6970 PowerTune Technology | Geeks3D

       

      Interestingly, Valley and Heaven showed no difference between -50 , +50 , or 0 (default). I am assuming that is because these benchmarks ignore the driver/CCC settings.

       

      More here: AMD graphics performance

        • Re: Catalyst Power Limit settings
          black_zion

          Left is -50% power, middle is 0% power, right is +50% power.

          Clipboard.jpg

           

          -50% yields minimum -50%, max -50%, and average -52%

          +50% yields minimum +17.5%, max +16.67%, and average +17.9%.

           

          I run 0% while gaming because I either get 60FPS, or I'm CPU bound, so it makes no difference if I use 50%.

           

          What was your resolution?

            • Re: Catalyst Power Limit settings
              kingfish

              I ran it at 1920x1080 and these are my 3D settings:

               

              with overdrive games.pngScreenshot_3.png

                • Re: Catalyst Power Limit settings
                  black_zion

                  Ah, here's the same 3 tests (-50, 0, +50) at the 1920x1080 preset benchmark. Percentages are about the same.

                  Capture.PNG

                   

                  My settings are the same as yours except for AA filter is Edge Detect, and Surface Format Optimization is On.

                    • Re: Catalyst Power Limit settings
                      kingfish

                      I think the most impressive increase is the minimum fps...that can make or break a game...

                        • Re: Catalyst Power Limit settings
                          black_zion

                          True with two asterisks. One being that if the minimum is still above your monitor's refresh rate and you are using Vsync to maintain a constant FPS figure, or above the FreeSync/GSync range if you have a capable monitor, then it's still going to be smooth. The second is that if the minimum happens during rare instances such as intense explosions, a drop for a second or two isn't going to break playability.

                            • Re: Catalyst Power Limit settings
                              kingfish

                              What I mean is, a 17% increase in the minimum frame rate has got to make a difference playing a demanding game with a marginal graphics card,

                                • Re: Catalyst Power Limit settings
                                  black_zion

                                  Capture.PNG

                                  Something odd which makes me question Furmark. I dropped my overclock back to the factory clock of 1020/1350, and did a test at power limit 0 (left) and 50% (right). While the +50% test shows a a difference of about -5% from the previous test, which is to be expected as my overclock is about 5.56% over stock, the 0% test is identical (I would say the 67fps reading of my overclock test is an artifact). I imagine it is because Furmark slams cards more than games do. So I did 3 tests in 3DMark11 (I do not have a license for the latest version, and there's no way in Hell I'm paying $20 for it): Stock clocks and power limit +50% (top), Overclock with power limit 0% (middle), and Overclock with power limit +50%. Both the overclock tests score identical numbers, with the overclocked tests scoring 5.3% higher than the stock clocked test.

                                   

                                  Clipboard.jpg

                                   

                                  I wish one of the reputable review sites would do an article on this though to settle it once and for all, as they have access to 5ghz i7s, a multitude of games, repeatable tests, and more patience than we do.

                                    • Re: Catalyst Power Limit settings
                                      Thanny

                                      With an absurdly unrealistic "benchmark" like Furmark, which is designed to max out a GPU's temperature (i.e. maximize power consumption), I'd expect to see power limit adjustments make a difference.

                                       

                                      With a more realistic load, I would expect the card's computational resources to be exhausted before the normal limit on power consumption is reached.

                                       

                                      There may be a handful of games which see some miniscule benefit from a higher power limit, but they are probably few and far between.  Which means even review sites probably aren't a good choice to find them - they always test a small set of games, usually including a couple older ones that have become benchmarking mainstays, and a sampling of new and popular ones from the past year.

                                       

                                      While Hawaii or later is probably a better test bed, I checked out a few games and benchmarks with my sole surviving 7970, and couldn't find one which caused it to downclock at a power limit of +0.  That's with both the factory 1000MHz and my standard 1200MHz overclock.  Everything downclocked at least a bit with the power limit set to -20.

                                        • Re: Catalyst Power Limit settings
                                          black_zion

                                          Same here, none of my games cause downlocking. However it must be said that the power limit would have a greater effect with AMDs really crappy stock air coolers than third party ones. kingfish always recommends setting the limit full up, while I always recommend leaving it alone, just wanted a more real-world case to see if it really did make any difference.

                                           

                                          That being said, I strongly prefer the older method of adjusting Vcore and VRAM voltages instead of the power slider.