With Novalogic's Comanche 4
I guess you could call Comanche 4 the old man on the
block, as it is probably the oldest of the tests in our
current suite of video card benchmarks.
Comanche 4 uses DX8 class
pixel and vertex shaders to produce some of the
realistic visuals used throughout the game. Unlike the
previous tests, this benchmark is heavily influenced by
CPU and system memory performance, especially at lower
resolutions. However, when the resolution is raised and
AA and Anisotropic filtering are enabled, the current
crop of video cards tend to slow down quite a bit
seen and reported in the past, the Comanche 4 Demo is much
more of a CPU-bound test than any of the other benchmarks we
have used. So much so, in fact, that at lower
resolutions we saw little difference between the Radeon
scores, although the GeForce FX 5600 still slipped to the
back of the pack. Applying anti-aliasing
affected the 5600 the most, dropping it by almost 60% when
using 6XAA. The two Radeons were able to handle this
much better with the 9600 XT losing at most 11 frames per
about 22%. At 1600x1200, the Radeon 9600 Pro and XT
models easily and thoroughly outpaced the GeForce FX 5600
until AA and AF were enabled, which really put the three
cards into equal footing. Again we see that AF is
having less of an effect on the 5600 benchmarks, as the 5600
and the 9600 Pro scores are very similar in the bottom graph.
Benchmarks / Comparison
Wolfenstein: Enemy Territory
Game, Better Effects, Old Engine
Territory can also be included in the "older" crowd,
although its focus is on OpenGL testing
(and yes, that means Quake 3 Arena is officially gone).
ET is a free standalone multiplayer game that is based
on the original Return to Castle Wolfenstein.
It uses a modified Quake3
engine yet exhibits plenty of CPU scaling and platform
variation, which also makes it a good benchmarking tool.
We used the built-in timedemo benchmark, with a customer
timedemo, which plays
back a recorded multiplayer session.
Comanche 4, Wolfenstein doesn't show too much of a disparity
when swapping video cards, with the 5600 almost beating out
the 9600XT at 1600x1200. When we started enabling AA
in the control panels, the 5600 finally earned the top spot,
beating both Radeons pretty handily. It seems NVIDIA
still handled OpenGL quite well, and although we could not get 6XAA to
work with this benchmark we feel confident that the 5600
would have still come out on top. We saw another anomaly
when trying to run 4XAA and AF at 1600x1200 on the 9600XT.
Although the 9600 Pro ran fine, the 9600XT seemed unable to
accept these values, and actually posted a number higher
than the non-AA score. This might just be a
driver issue that needs to get ironed out with the new XT
line of Radeons and Catalyst 3.8 drivers.
X2 The Threat, Overclocking & Our Final