Unreal Tournament 2003 (UT2003) has become a popular benchmark for measuring the performance of today's graphics cards. Developed by Digital Extremes and Epic Games, the graphics in UT2003 were developed using Microsoft's DirectX applications programming interface. While UT2003 contains a comprehensive suite of benchmarking features, there's no substitute for measuring "real-world" performance other than by playing the game. Plain and simple - and fun!
Although UT2003 is primarily targeted as a multi-player game, it has a mode that allows the player to create matches against computer controlled bots, otherwise known as bot-matches. It's under this scenario that I put my system up to the test against one of the most demanding games on the market.
The demo of UT2003 contains two levels, or maps, that serve as your battlefield. The outdoor map DM-Antalus will cause greater stress on your computer system then DM-Asbestos. Outdoor levels typically contain a much larger field of view, which results in an increase in the number of polygons and textures that are processed by the graphics hardware.
GAME PLAY CONFIGURATION
The following sections list the graphics and sound menu settings in UT2003 that were used during game play sessions.
Full Screen - 32-Bit Color
Texture Detail - Normal
World Detail - Highest
Character Detail - Normal
Physics Detail - Normal
Character Shadows - Enabled
Dynamic Lighting - Enabled
Detail Textures - Enabled
Projectors - Enabled
Decals - Enabled
Coronas - Enabled
Decal Stay - Normal
Foliage - Enabled
Trilinear Filtering - Enabled
Coronas - Enabled
Audio Mode - Hardware 3D Audio
Message Beep - Enabled
Mature Taunts - Enabled
Play Voices - All
Announcements - All
The following sections contain a listing of my system specifications.
AMD Athlon XP 2700+ @ 2.17GHz - Thouroughbred Revision B
Windows XP Professional with Service Pack 1 / DirectX 8.1
Expert System Performance
166MHz Front Side Bus
Memory Timing 4-2-2
CAS Latency 2.0
APIC Mode Disabled
Frame rate data was collected using FRAPS, which runs as a background task. Two deathmatch games were played against six "skilled" bots with each match lasting around three minutes. The benchmark results are based on averaging the frame rates obtained from the two matches.
Having run through a number of test cases with UT2003 and bot-matches, the minimum and average frame rates reported by FRAPS typically varied 1-3 frames per second. The maximum frame rate experienced greater variations.
GAME PLAY RESULTS
Performance is shown with bar charts that display the maximum (red), average (green), and minimum (blue) frames per second, per resolution. The first charts show my results with no antialiasing and trilinear filtering from the outdoor level DM-Antalus.
Playing DM-Antalus at resolutions of 1024x768 and 1280x960 offered acceptable performance in each match. Although 1280x960 failed to reach an average frame rate of 60, a minimum frame rate of 30 is an indication that noticeable slowdowns in performance didn't occur. I could increase performance by 10% if player shadows were disabled.
Let's move on to the less demanding indoor map DM-Asbestos.
If UT2003 only consisted of indoor maps, I might be able to con myself into playing at a resolution of 1600x1200! But since a player typically cycles through a number of indoor and outdoor maps, the resolution used will be a reflection of the results from an outdoor level like DM-Antalus. Keep in mind that increasing performance can easily be done in real-time by using a key that's been configured as an on/off toggle. For example, this key could be used to turn off player shadows prior to entering a level.
I'll continue to use the outdoor DM-Antalus map and also enable 2X antialiasing.
The best I can do with on the GeForce4 Ti 4600, while maintaining a playable frame rate, is 2X antialaising at a resolution of 800x600. An average of 44 frames per second at 1024x768 is a bit too sluggish.
And in conclusion, the DM-Antalus map with 2X anisotropic filtering.
With 2X anisotropic filtering, the results are about the same as with 2X antialiasing, and achieving playable frame rates will be limited to a resolution of 800x600.
I hope you enjoyed looking over this brief article on measuring graphics performance based on actual game play. Benchmarks can be run on your favorite games with FRAPS and can reveal information about your system you may not have known otherwise.
Oh yea, my favorite settings when playing UT2003 - 1280x960 with no antialiasing or anisotropic filtering :)