The Test

As of yet, ATI does not have a comparable part to the 6600 or the 6600 GT. They are working on bringing an R4xx based midrange chip out at some point in the future, but we haven't gotten any concrete details about possible parts yet. For this review, we took a look at the X600 XT as a comparison point, but as it turns out, this isn't as helpful as we had hoped. The major problem is that we wanted a good comparison with current and previous generation GPUs, and the only way we could effectively do that was by comparing AGP cards to these PCI Express solutions. As such, most of the numbers other than the 6600 GT and the X600 XT have been run on our 3400+ graphics system. The Doom 3 and Source Engine AGP benchmarks were run on the overclocked FX53, and so aren't as comparable to the rest of the world.

But we feel that this is an acceptable setup in light of the fact that PCI Express systems will have to compete with AGP systems. Our graphics test platform is based on AMD, and this serves us well enough in simply comparing how well the 6600 does in modern games (which are mostly graphics limited anyway). As long as the processor is powerful enough to keep from becoming a large bottleneck we will have a clue about NV43 performance. And we feel we've accomplished this. Here's our test setup.

Performance Test Configuration

Processor(s):

AMD Athlon 64 3400+
AMD Athlon 64 FX53 (oc to 2.6GHz)
Intel Pentium 4 3.4GHz EE

RAM:

2 x 512Mb OCZ 3500 Platinum Ltd (2:3: 2:10)
2 x 512MB Micron DDR2 533

Hard Drives

Seagate 120GB 7200 RPM (8MB Buffer)

Video AGP & IDE Bus Master Drivers

VIA Hyperion 4.51
Intel Chipset INF 6.0.1.1002

Video Card(s):

NVIDIA GeForce 6600 GT
NVIDIA GeForce 6800 Ultra Extreme
NVIDIA GeForce 6800 Ultra
NVIDIA GeForce 6800 GT
NVIDIA GeForce 6800
NVIDIA GeForce FX 5950
ATI Radeon X800 XT Platinum Ed.
ATI Radeon X800 Pro
ATI Radeon 9800 XT
ATI Radeon 9700 Pro
ATI Radeon X600 XT

Video Drivers:

ATI Catalyst 4.6
ATI Catalyst 4.8
NVIDIA ForceWare 61.77
NVIDIA ForceWare 65.37

Operating System(s):

Windows XP Professional SP2

Motherboards:

MSI MS-6702E (VIA K8T800 Pro Chipset)
FIC K8T800
Intel D925XCV



The 61.77 drivers were used on all but the 6600 GT, which was powered by the 65.37 beta drivers. On the ATI side, most of the numbers shown here were run with the 4.6 Catalyst, except the X600 XT which was run with the 4.8 version.

High-Tech Mid-Range Aquamark3 Performance
Comments Locked

44 Comments

View All Comments

  • ViRGE - Tuesday, September 7, 2004 - link

    Carfax, they have a standard testbed, so the numbers from the other NV cards come from previous benchmarks. If they upgrade the drivers on everything else, they'll have to re-run the benchmarks on everything else.
  • Cybercat - Tuesday, September 7, 2004 - link

    Very nice performance. Best mainstream card in a LONG time.
  • Carfax - Tuesday, September 7, 2004 - link

    WHY where two sets of drivers used?!? Why couldn't you just use the 65.76 drivers for both the 6600GT and the rest of the Nvidia cards?
  • ViRGE - Tuesday, September 7, 2004 - link

    And continuing on #20's tangent, the 5950 Ultra beats the 6600GT by 200%? That definitely isn't right.
  • railer - Tuesday, September 7, 2004 - link

    I don't think those Jedi Knight results are correct. 9800xt beats the 9700 pro by 300%? I think not......
  • Doormat - Tuesday, September 7, 2004 - link

    "That isn't to say that they are less power hungry than an AGP card that requires external power, but that the PCI Express slot supplies enough voltage to the card that it doesn't need any more juice."

    Nitpick: the AGP and PCIE slots provides enough voltage, but the main restriction is current. Each spec is designed to deliver so many amps of current at the specified voltage. As cards get bigger and badder, they draw more current, and need the extra power hookups.
  • LoneWolf15 - Tuesday, September 7, 2004 - link

    I'm puzzled as to why the 6600GT beats the 6800(straight) so often. Doesn't the 6800 have a full 256-bit path?
  • neogodless - Tuesday, September 7, 2004 - link

    Yes, fix "looses" on the final page!

    Crazy to see a Radeon 9700 Pro do so poorly... very surprised it's doing poorly compared to the 9600/X600. Is that right? Doesn't seem right to me...
  • Jalf - Tuesday, September 7, 2004 - link

    #14: Yeah, but looks kinda weird that half the charts only shows those two cards, while the other half shows the full spectrum :)
  • Falloutboy - Tuesday, September 7, 2004 - link

    looks pretty good to me even at its worse its still on par with a 9800xt and in alot of games is besting the x800pro looks like a pretty good deal at 200 bucks prolly will go cheaper once it hits the stores in mass

Log in

Don't have an account? Sign up now