Announcement

Collapse
No announcement yet.

NVIDIA GeForce Power Efficiency: From The 6600GT To The GTX 750 Ti

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

  • NVIDIA GeForce Power Efficiency: From The 6600GT To The GTX 750 Ti

    Phoronix: NVIDIA GeForce Power Efficiency: From The 6600GT To The GTX 760 Ti

    When NVIDIA was doing their press briefings for their new Maxwell architecture they frequently talked up its power efficiency and how the power efficiency is four times greater than where it was four years ago with Fermi... But how is Maxwell and NVIDIA's power efficiency compared to hardware from ten years ago? In this article we have done fresh benchmarks -- with power consumption, thermal, and performance-per-Watt measurements -- of NVIDIA's mid-range graphics cards from the week-old GeForce GTX 750 Ti to as far back as the GeForce 6600GT (NV43) graphics card from 2004.

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Originally posted by phoronix View Post
    Phoronix: NVIDIA GeForce Power Efficiency: From The 6600GT To The GTX 760
    I was getting excited that you had a scoop on a GTX 760 and then had my hopes dashed when I realised it was a typo.

    Comment


    • #3
      Interesting, so the GT 520 wins in sheer power efficiency under load, with the GT 220 not far behind. Although I'd still be interested in idle power usage.

      Comment


      • #4
        wtf happened to the 9600gso? it should be between 8600gt and 8800gt and more near the 8800gt

        Do you have some kind of crippled DDR2 version? or is it a driver regression?

        Still quite content with my gtx460^^

        Comment


        • #5
          Originally posted by Herem View Post
          I was getting excited that you had a scoop on a GTX 760 and then had my hopes dashed when I realised it was a typo.
          The GTX 760 already exists and it's a Kepler part: https://en.wikipedia.org/wiki/Compar...rce_700_Series

          Comment


          • #6
            Originally posted by GreatEmerald View Post
            Interesting, so the GT 520 wins in sheer power efficiency under load, with the GT 220 not far behind. Although I'd still be interested in idle power usage.
            Ehh not really, it takes less power which is true, but it can't do the same as more powerfull card to call it energy efficient(what the hell is power efficiency anyway). Pity not to see kepler based gt630 on these tests, it would be rather efficient too with it 25W tdp.

            Comment


            • #7
              Originally posted by rudl View Post
              wtf happened to the 9600gso? it should be between 8600gt and 8800gt and more near the 8800gt

              Do you have some kind of crippled DDR2 version? or is it a driver regression?

              Still quite content with my gtx460^^
              There were two different 9600gso boards. One was G92 based and the other was G94 based. The former is the earlier one and performed much better on most loads--96 CUDA 'cores' will do that when compared to the 48 of the G94 chip. The only thing in the favor of the G94 was the later G94b die shrink (for power savings) and the 256 bit memory interface with the 16 ROPs that went with it. For some fill rate limited loads, the G94 was a better card. I have one of both and it's about a 2 to 1 performance difference in real games and F@H.

              So, unless you know which of them is being benchmarked, a 9600GSO is a useless datapoint.

              Comment


              • #8
                Originally posted by tuke81 View Post
                Ehh not really, it takes less power which is true, but it can't do the same as more powerfull card to call it energy efficient(what the hell is power efficiency anyway). Pity not to see kepler based gt630 on these tests, it would be rather efficient too with it 25W tdp.
                Yes, but for things like HTPCs it's enough power as it is, so the energy use itself is most important in such cases (and also idle energy use as well).

                Comment


                • #9
                  Originally posted by GreatEmerald View Post
                  Yes, but for things like HTPCs it's enough power as it is, so the energy use itself is most important in such cases (and also idle energy use as well).
                  Yeah that's true, more power more heat to dispatch. Though this new maxwell architecture introduced new power state called ?GC5? specifically for low usage tasks such as video playback. Which one can really see techpowerup tests(gtx750ti idle 4W,multimonitor 5W, bluray 6W vs gt520 idle 6W, multimonitor 9W, bluray 10W):
                  Today, NVIDIA launched their GeForce GTX 750 Ti, which is based on the new GM107 graphics processor introducing NVIDIA's Maxwell graphics architecture. Power consumption of the new card is at a record low, which means temperature and noise levels are impressive as well.

                  ZOTAC's GeForce GT 640 is based on NVIDIA's new Kepler architecture. Thanks to its energy efficient design, ZOTAC managed to release a single-slot graphics card that will be interesting to many media PC builders. It also comes with HDMI and 2x DVI outputs, adding to the card's versatility.

                  Comment


                  • #10
                    Originally posted by willmore View Post
                    There were two different 9600gso boards. One was G92 based and the other was G94 based. The former is the earlier one and performed much better on most loads--96 CUDA 'cores' will do that when compared to the 48 of the G94 chip. The only thing in the favor of the G94 was the later G94b die shrink (for power savings) and the 256 bit memory interface with the 16 ROPs that went with it. For some fill rate limited loads, the G94 was a better card. I have one of both and it's about a 2 to 1 performance difference in real games and F@H.

                    So, unless you know which of them is being benchmarked, a 9600GSO is a useless datapoint.
                    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite
                    Michael Larabel
                    https://www.michaellarabel.com/

                    Comment

                    Working...
                    X