Performance Test: Unreal Tournament 2003, CTF-Face3

This content was originally featured on and has been converted to PC Perspective’s website. Some color changes and flaws may appear.

Let’s now take a look at performance in Unreal Tournament 2003. With the Unreal Tournament 2004 “expansion” coming out very soon, it’s even more important now to see how current cards run with this game. Unreal Tournament 2004 features “large-scale maps … awesome new weapons and land- and air-based vehicles“, so the CTF-Face3 benchmark is becoming more important.

Unreal Tournament 2003. CTF-Face3.


  • Ultra-High details
  • 1024×768 and 1600×1200 resolutions
  • Using 60FPS as the “playable” mark. This gives enough breathing space for a real match with other players.

Unreal Tournament 2003. CTF-Face3. 1024x768 benchmark results.
* FX5600 set to 6XS anti-aliasing. **Playable FPS: ~60 FPS or higher.

Unreal Tournament 2003. CTF-Face3. 1600x1200 benchmark results.
* FX5600 set to 6XS anti-aliasing. **Playable FPS: ~60 FPS or higher.

General Remarks
Be aware that the performance of the FX5600 may be inaccurate. In the weeks leading up to the publishing of this article, there have been numerous discussions about issues with the GeForce FX that may affect UT2K3 performance. So don’t base your purchasing decision on a UT2K3 benchmark alone. You should take into account all possible benchmarks to get the best idea of how a particular card performs.

Radeon 9700
The Radeon 9700 does really well in the UT2K3 test. The fact it can perform the CTF-Face3 benchmark at 1024×768 without any problems is a big achievement. At 6×16, you see it dip well below the 60FPS playable mark.

At 1600×1200, 4xAA and 0xAF is the highest playable setting with Ultra-High details, but you can probably get further by taking down in-game details.

Radeon 9500
The Radeon 9500 has a strong showing in CTF-Face3. It can perform reasonably well at 6xAA with some aniso-tropic filtering enabled. 16xAF on any anti-aliasing setting appears somewhat out of reach unless details are turned down.

The only playable setting at 1600×1200 is 0x0. At any higher setting we see performance below what many of us consider unacceptable. Interesting to see that initially the Ti4800SE beats the Radeon 9500, but the Radeon 9500 comes out on top in 0x8 tests and higher (even though the FPS is unplayable at that point). More on this in the Ti4800SE remarks below.

The FX5600 has a very strong showing up to 4×8 AAxAF at 1024×768. At 4×8, it pulls off an impressive 71.7 FPS which places it between the Radeon 9500 and the Radeon 9700. We can see that whenever anisotropic filtering is turned on, the FX5600 out-performs the Radeon 9500.

At 1600×1200, we see that the FX5600 is able to make a reasonable showing for the 0x0 test with 60 FPS. We see the performance degrade at each successive test. But since anti-aliasing is not as important at higher resolutions, you should be able to get some decent frame rates with 0x8 with some details turned down.

The FX5200 pulls off some respectable numbers up to and including 2xAA at 1024×768. Given the price of this particular card, it’s not too surprising that it doesn’t perform too well at higher details and at higher resolutions. However, it is great to see a budget card capable of playing CTF-Face3 with 2xAA enabled.

The Ti4800SE has a decent showing up to 4x0AA at 1024×768. At 1600×1200, we see something pretty drastic. The performance drops nearly exponentially from 82.0 FPS to 31.5 FPS to 13.9 in three successive tests. Unlike all other cards in this review, the Ti4800SE degrades considerably at high resolution (1600×1200). This kind of behaviour is not seen on any other test (i.e. Freelancer or Morrowind). It makes me curious to find out exactly what is causing this.

« PreviousNext »