New Benchmark Suite
Here're the benchmarks and games we are using to determine the performance of the new ATI Radeon XT:
- Unreal Tournament 2003 v2225* : We solely use our own-recorded timedemos of the UT2003 maps Inferno and Magma.
- Wolfenstein - Enemy Territory : We're using a replay recorded during a clan war.
- AquaMark 3 : Measuring Triscore and other values
- Tomb Raider - Angel Of Darkness v49* : Using a own recorded timedemo in the level Paris1c
- X2 Rolling Demo
- Command & Conquer Generals Demo* : Replay of a self recorded game and measured with FRAPS. We disable the framerate limiter by hacking the gamedat.ini included in the file INI.big.
- Warcraft III - The Frozen Throne : Using a replay recorded during a clanwar (3on3) on the map "Bloodstone Mesa". We run the replay at 8x speed to stress the graphics as much as possible and measure the framerate with FRAPS.
- Freelancer : Replay of a cutscene, measuring the framerate with FRAPS
- MS Flight Simulator 2004 : Using a self recorded flight and measuring the framerate with FRAPS
- Battlefield 1942 - Secret Weapons of WWII Demo* : Running a certain path in the game by hand. Measuring the framerate with FRAPS.
* This game is part of NVIDIA's "The Way It's Meant To Be Played" program.
Since the actual NVIDIA Detonator driver v45.23 has issues in some applications (see: Aquamark3: Accurate Benchmarking for Old and New (DirectX9) Apps? ), we also use the older driver version v44.03. The new Detonator 50 alias ForceWare driver was not yet available at the time of this review. NVIDIA asked us not to use the beta version v51.75 of this driver in reviews because it contains a couple of bugs. All ATI cards were tested with Catalyst drivers v3.7.
The result tables from benchmarks with build-in framecounters are in blue color. The results measured with the framecounter tool FRAPS are in gray color. Gameplay results, where we manually play a game in the same way by running the same path in all tests, are in green color.
Since the new NVIDIA drivers are not available yet and we're also waiting for the new ATI Catalyst v3.8 release, it does not make much sense to take a closer look at the image quality at this time, since there are no changes of what we saw in the weeks before. Please have a look at the following articles to get an overview of what we have today:
- Aquamark3: Accurate Benchmarking for Old and New (DirectX9) Apps?
- NVIDIA Detonator FX 45.23
- Introducing ATi's New Middle Class: The Radeon 9600 PRO
- NVIDIA GeForceFX 5900 Ultra: The Way FX is Meant to be Played!!
- Strike Force: The new ATI Radeon 9800, 9600 and 9200 Series
|Motherboard||ASUS P4P800 Deluxe, Intel i865PE chipset|
|Processor||Intel Pentium 4 3.2 GHz (FSB800)|
|RAM||2x 256 MB Kingston HyperX DDR400CL2 / 6-2-2-2|
|Drivers & Configuration|
|Graphic Cards||ASUS Radeon 9800 XT 256 MBATI Radeon 9800 XT 256 MB (Reference)ATI Radeon 9800 Pro 128 MB (Reference)ATI Radeon 9700 Pro 128 MB (Reference)ATI Radeon 9600 Pro 128 MB (Hercules)ATI Radeon 9500 Pro 128 MB (Reference)NVIDIA GF FX5900 Ultra (Reference)NVIDIA GF FX5600 Ultra FC (Reference)|
|Graphics Driver||Detonator FX v44.03Detonator FX v45.23ATI Catalyst v3.7|
|OS||Windows XP Professional SP1|
Unreal Tournament 2003
UT 2003 is a well-known last-generation first-person shooter. Simply put, the 3D engine is an extended DirectX 7 engine with a few DirectX 8 features added on. In some cases, like terrain rendering, the engine already uses pixel shaders v1.1 or v1.4, depending on hardware support. The "Unreal Engine" is under constant development, always being improved and tweaked. Over the last few years, several games using various versions of this engine have been published - see wiki.beyondunreal.com for more info.
We ran this benchmark using the highest detail settings available to put as much strain on the graphics capabilities as possible. Instead of using the build-in Flyby demos, which have very little to do with the real performance while playing the game, we recorded our own timedemos in the maps "Inferno" and "Magma". The interesting detail is that both maps have different demands on the cards as you can see in the 4xFSAA&8xAniso results.
Using no FSAA and anisotropic filtering, the FX5900 and Radeon 9800 Pro are at the same level. The XT is only a bit faster but leading the pack.