The following configuration represents the testbed used throughout the review process.
AMD Athlon XP 1600+ (Unlocked, though operating at default 133MHz FSB)
MSI KT4 Ultra KT400 Motherboard
(2) 256MB Corsair XMS PC3200 CAS2
Western Digital WD800JB 80GB 8MB HDD
Chieftec Aluminum Case With Deer 430W Power Supply
Sapphire Radeon 9700 Atlantis Pro
Gateway VX1100 21-Inch CRT Monitor
Windows XP Professional / DirectX 8.1
ATI 220.127.116.1193 Beta Drivers
VIA 4-in-1 443 Chipset Drivers
Rage3D Tweak Version 3.6
Athough the Athlon XP 1600+ has been successfully unlocked, I chose to run the processor at default clock speeds. While this setting becomes a performance bottleneck, the benefits outweigh the consequences. Far too often test configurations use the latest and most expensive processors available. The fact of the matter is, not everyone can afford or has access to this hardware. Using a more common processor speed, the benchmarks will give us an accurate depiction of how a standard system will perform with the latest generation of high-end graphics cards.
In order to attain a better feel for how the Radeon 9700 Pro performs, a diverse selection of games were chosen for testing. The following software was used in this review for benchmarking purposes.
3DMark2001 SE Build 330
Quake 3 Arena Version 1.32
Unreal Tournament 2003 Version 1080
Jedi Knight 2 Version 1.04
Battlefield 1942 Version 1.1
Need For Speed Hot Pursuit 2 Demo
No One Lives Forever 2 Demo
Given the fact that only a handful of the selected applications contain a built-in benchmarking feature, I also relied on version 1.9 of FRAPS to measure performance. In doing so, I made certain that each benchmark test used the same constraints and was accurately measured. In addition, this method of testing provides realistic gameplay scenarios in lieu of fixed loop scenerios, which are used in traditional benchmarks. One exception of this methodology is in use of HardOCP's UT2K3 Demo Benchmark version 1001.
Benchmark results are reported at various resolutions using three different image quality levels. The initial tests were conducted with standard image quality settings with neither antialiasing nor anisotropic texture filtering enabled. The second test added 6X antialiasing, while the final test included 6X antialiasing along with the maximum level of anisotropic texture filtering, which is 16X.
Direct3D Image Quality Settings
Despite the adverse effect enhanced image quality settings can have on overall system performance, the visual improvements are often desirable. The screenshot below illustrates this fact perfectly. As it is, the screenshot represents an image from UT2003 taken at a resolution of 1600x1200. Clicking on the image will cause a larger image to be shown. By moving the mouse over the larger image, we are presented with a screenshot taken at the same resolution with a few image quality enhancements. Here, the maximum settings for antialaising and anisotropic texture filtering are used to make the image more crisp and realistic.