|
Although the PCB
itself adheres to NVIDIA's GeForce FX 5900 Ultra reference
design, Asus took some steps to ensure the V9950 Ultra is
easily discernable among its competition. After a long
stretch in '02 of reviewing one GF4 Ti after another, with
only a sticker applied to a fan to tell them apart, we were
glad to see Asus make some changes that really help
differentiate this card from other GFFX 5900 Ultras.
The V9950
Ultra's blue PCB was the first trait to catch our eye.
A colored PCB does nothing for performance, but it does help
give the card some style and distinguish it from the
plain-vanilla green PCBs being sold be some other
manufacturers. To give the V9950 Ultra even more
personality, Asus also uses a gold-plated external plate and
gold-plated DVI and DB15 connectors. Asus claims the
plating improves signal quality, but in side-by-side
comparisons with an NVIDIA reference card and an Abit 5900
non-Ultra, we didn't see any image quality differences in
either 2D or 3D (up to 1600x1200x85Hz). The V9950
Ultra's distinguishing factor has got to be its custom
copper cooling solution. The NVIDIA designed,
dual-slot FX-Flow cooler is gone. Instead, Asus
employs a specially designed, quiet 2-Fan copper heatsink on
the V9950 Ultra, that no only does an excellent job cooling
the GPU and RAM, but it occupies only a single slot!
We pulled the cooler off the card and found it to be very
well mated to the board's components. The heat plates
mounted to the front and back of the card made perfect
contact with the NV35 GPU and 256 RAM (16 - 2.2ns Hynix BGA
chips), although we would have liked to have seen a bit more
thermal paste applied to the GPU. Like other GeForce
FX 5900's, the fans on the V9950 Ultra spin-up or down when
switching from 2D to 3D mode. Even when running at
full speed, the fans used on the this card are relatively
quiet. We could not hear the V9950 Ultra over the CPU
and case fans used in our test system.
If you have
inspected a few GeForce FX 5900s, you may have noticed one
omission from the V9950 Ultra's PCB. If you take a
look at the last picture above, at the lower left corner
you'll see a bare spot on the board (here's
a shot of Leadtek's A350 PCB for comparison). That
location is usually reserved for a video encoder of some
sort. The V9950 Ultra that we're looking at today does
not have any Video-In capabilities, which is why an encoder
is not present. Knowing Asus, however, we wouldn't be
surprised to see a "Deluxe" version of this card with ViVo
functionality introduced sometime in the near future.
|
|
|
Screenshots While Gaming With Gun Metal |
|
Yes,
It's a DX9 Game! |
|
1024x768 - 4X AA
Before we sat
down to benchmark the V9950 Ultra, we spent some time gaming
with this beast. This was actually a good month for
gamers. Some highly anticipated games and demos were
all released recently. We played with the
Call of Duty demo,
Tron 2.0,
Gun Metal and
Unreal Tournament 2003. All of the games ran very
well on the V9950 Ultra, even with anti-aliasing and
Anisotropic filtering enabled (which is to be expected
with current titles running on a $450+ dollar card!).
We snapped off a few screenshots from Gun Metal, one of
the games included with the V9950 Ultra, to give you some
idea of the kind of in-game images this card is capable of
displaying. All of the shots above were taken at
1024x768 with 4XAA and 8X Anisotropic filtering enabled.
|
|
|
Screenshots with Antialiasing Enabled |
|
Works
Great, But With Some Weirdness |
|
1024x768
NO AA |
1024x768
2X AA |
1024x768
4X AA |
1024x768
4XS AA |
1024x768
6XS AA |
1024x768
8X AA |
We also took a
few screenshots to specifically show you the benefits of the
GeForce FX 5900 Ultra's Antialiasing techniques, using the
exact same frame from 3DMark03's "Wings of Fury" demo.
As you can see, as the AA level is increased, the visible
jaggies in the image are significantly decreased. Pay
special attention to the small plane in the background at
the upper-left of the images. Open up a few of the
images, and switch between them quickly to really see the
benefits of anti-aliasing. We did notice something
strange when using any level of AA higher than 4X. The
smoke in this scene is rendered differently at 4XS, 6XS and
8X AA levels. The sampling pattern used and filtering
applied to the image when using these modes is probably what
causes this anomaly, so sticklers for image quality should
probably stick with 4XAA (or lower).
Let's Start Benchmarking
|
