Geforce 8800 GTS 320Mb Review - Does 640Mb matter?

Videocards/VGA Reviews by geoffrey @ 2007-02-12

With the introduction of the Geforce 8800 GTS NVIDIA made available an affordable high end DX10 video card, now they´ve slashed its memory in half, reduced the price, to make it even better deal. Can the new GTS keep up with its memory-loaded brother? We find out.

  • prev
  • next

Test setup & Benchmark methodology

Test setup

Geoffreys' Intel Test Setup
Madshrimps (c)
CPUIntel E6600 @ 3,6GHz
CoolingZalman 9700 LED
MainbordAsus P5B Deluxe Wifi
Memory2x1Gb TEAMGROUP Xtreem 800MHz 4-4-4-10
Other
  • OCZ Powerstream 600W PSU
  • Maxtor 80Gb PATA HDD
  • Seagate 200GB SATA HDD
  • 20" Dell UltraSharp 2007FP TFT monitor

  • The CPU was running at 3,6GHz by setting the front side bus to 400MHz and keeping the multiplier at default (9). The memory was running @ 400MHz (800MHz DDR) with 4-4-4-10 timings 1/1 with the FSB.
  • ForceWare 97.92 drivers
  • While Windows Vista is now officially launched, and G80 are the only DX10 capable cards for now, we decided to test with a mature Windows OS (XP SP2), even if we wanted to, the lack of none-beta drivers for Vista keeps us from testing on the new platform.

    Benchmark methodology

    We compared both versions of the 8800GTS and try to find out how the amount of memory affects performance.

    All tests were done with a 20” LCD monitor with maximum resolution of 1600*1200. For high-end VGA cards this resolution can be quite stressful in newer games especially when higher quality settings are enabled like anti aliasing and anisotropic filtering.

    The G80 core is powerful, at 1280x1024 and lower resolutions we couldn’t properly stress them unless we cranked the AA levels beyond reasonable levels. We stuck at our monitor’s maximum native resolution of 1600*1200 and found a good balance between image quality and playability.

    Depending on what game we ran, IQ was set a bit different. You will be informed at all time at what it is actually set. Whenever possible the in-game quality options were set to highest. FRAPS was used to measure the FPS during repeated manual run-throughs of a certain part of the games tested, the minimum, maximum and average values were recorded.

  • FEAR
  • Battlefield 2
  • Tomb Raider: Legend
  • Oblivion

    We also ran Futuremark's 3D mark software and overclocked our VGA cards using the latest version of Rivatuner, which is also used to log heat/clocks. Noise is being recorded with the SmartSensor AR824 digital sound level meter.
    • prev
    • next
    Comment from Rutar @ 2007/02/12
    going right after [H] last stronghold with those graphs done with THE SAME SETTINGS, evil evil geoffrey


    I would like even more graphs of the AA tests because that's what matters not the number.
    Comment from jmke @ 2007/02/12
    we've been using squiggly charts since December 2005
    http://www.madshrimps.be/?action=get...&articID=4 13
    Comment from Sidney @ 2007/02/12
    Nice article
    Comment from Wolf2000me @ 2007/02/13
    300$ for a 320mb 8800GTS is tempting. Well worth considering although I'd prefer to wait for ATI's R600 cards. If those fit in my case, that is

    Although it's a nice article there are a few spelling errors. Namely in the specs table here:
    http://www.madshrimps.be/?action=get...&articID=5 43

    Frequency is spelled in three different ways of which only one is the correct one.
    Comment from geoffrey @ 2007/02/13
    Quote:
    Originally Posted by Wolf2000me View Post
    300$ for a 320mb 8800GTS is tempting. Well worth considering although I'd prefer to wait for ATI's R600 cards. If those fit in my case, that is

    Although it's a nice article there are a few spelling errors. Namely in the specs table here:
    http://www.madshrimps.be/?action=get...&articID=5 43

    Frequency is spelled in three different ways of which only one is the correct one.

    Oops

    Fixed it, thx
    Comment from jmke @ 2007/02/13
    couldn't get to the spelling errors at noon, sorry

     

    reply