Unreal Tournament 2004 Performance
Publisher: Epic Games, Inc. 
Minimum System Requirement:
Mac OS X: 10.2.8 and higher 
CPU: G4 @ 933 MHz 
RAM: 256 MB 
Hard Disk: 6 GB 
Graphics: 32 MB VRAM
RAM: 256 MB Hard Disk: 6 GB Graphics: 32 MB VRAM__________________________________________
Unreal Tournament is a favorite from way back.
It has always been known to be more CPU dependent than GPU dependent, but a graphics card can make a difference.
It is interesting to see just which card at what setting helps performance the most.
The first set of tests uses the Flyby map in SantaDuck.
Flyby tests will exercise the graphics card more than Botmatch tests, which are predominately CPU bound.
One note- Performance will vary with the different maps.
Just because 80+ fps is gained here, that doesn’t mean the game will run at 80+ fps.
Some maps are faster (well over 100 fps), and some are slower.
This is an illustration to show the impact a graphics card may have on performance.
Flyby: Sun Temple - 133 MHz bus G4 - OS X 10.5.6
G4 Quicksilver 2002 4x AGP Graphics
__________________________________________
Using the Botmatch tests, we see fps rates that will be closer to that of game play.
Again, test results will be faster or slower in different maps.
Though Unreal Tournament is a CPU bound game, graphics cards can have an impact on performance.
Botmatch: Asbestos - 133 MHz bus G4 - OS X 10.5.6
G4 Quicksilver 2002 4x AGP Graphics
__________________________________________
To illustrate how a different map will perform differently, compare the Asbestos match results (above) with the Antalus results (below). Antalus stresses the system a bit more than Asbestos.
Botmatch: Antalus - 133 MHz bus G4 - OS X 10.5.6
G4 Quicksilver 2002 4x AGP Graphics
__________________________________________
Well, you can see that, for the most part, graphics cards don’t make a big difference at lower resolutions.
In fact, the lower resolutions are usually as fast or faster with older cards.
Two reasons.
One, UT2K4 is a very CPU intensive game, rather than GPU intensive.
Two, in the case where a newer, more powerful card is actually slower than an older, slower card, this is because the graphics rendered are not the same.
The graphics cards tested can be divided into two groups.
The first, are the older cards that don’t have the later OpenGL support, and all rendered the graphics of UT2K4 the same way:
-
•Radeon 7500
-
•Geforce4 MX
-
•Radeon 9000 Pro
-
•Radeon 9700 Pro (actually in a group in between these two groups. Same OpenGL support as the 9800’s, but renders the same as the other cards in this group.)
The second group are all later generation, support OpenGL 2.0 100% and render graphics in a more detailed manner:
-
•Radeon 9800 Pro 128 and 256MB
-
•Radeon 9800 XT
-
•Geforce 6800 GT
-
•Geforce 7800GS
One thing that I noticed is the newer cards all render greater detail in the textures, though this is done with a cost to achieved fps.
Because of the greater detail when using a newer card, the system has to work harder to generate the graphics that are not generated with the older cards.
Another thing that is obvious, if the game is to be played at higher resolutions, a newer, more powerful graphics card is a must.
Whatever the monitor, playing any game at the native resolution of the monitor is desired for performance.
If you are like me, and want to run games on full screen with a large monitor (like my 23” Apple Cinema Display) a high performance graphics card is a must.
2.0 GHz 7447A G4 Quicksilver 2002 4x AGP Graphics - OS X 10.5.6
Asbestos Botmatch and Suntemple Flyby - Fullscreen 23” ACD
Here we can see the processor change provides for 5-10 fps in the Botmatch, and 5-17 fps in the flyby.
This type of performance increase is the difference between playable and enjoyable.
