Test Setup
| Basic Hardware | |
|---|---|
| Processor | AMD Athlon XP 2800+ (2.25 GHz) |
| Main Memory | 2x 256 MB PC3500 CL2, Corsair Micro |
| Hard Drive | IBM DeskStar 60GXP, IC35040, 40 GB, 7.200 U/ Min, NTFS |
| Graphics Card | Asus V8460, GeForce 4 TI4600, 128 MB |
| Common Hardware | |
| Network Adapter | 3COM 3C905-TX |
| Drivers & Settings | |
| Graphics Driver | Detonator 4 Series Version 30.82 |
| Chipset Drivers | nVIDIA driver 1.13 |
| DirectX Version | 8.1 |
| Screen Resolution | 1024x768x16, 85 HzSPECviewperf : 1280x1024x32, 85 Hz |
| Operating System | Windows 2000 Professional SP3 |
| Benchmarks | |
| Quake III Arena | Retail Version 1.16command line = +set cd_nocd 1 +set s_initsound 0Graphics detail set to ’Normal’Benchmark using ’Q3DEMO1’ |
| Unreal Tournament 2003 | Official Demo |
| Comanche 4 | Official Demo |
| WinACE | 2.20, best compression, Dictionary 4096 kB |
| SPECviewperf | Version 7.0, all tests |
| mpeg4 encoding | Xmpeg 4.5DivX 5.02 Pro (YV12)Compression/ quality : HighestData Rate : 780 kBitFormat : 720x576 Pixel@25 fps150 MB VOB file, no Audio |
| Sysmark 2002 | Internet Content Creation |
| Lame MP3 | Lame 3.91 MMX, SSE, SSE2, 3DNow ! |
| PCMark 2002 | Memory Bandwidth Benchmark |
| 3DMark 2001SE | Default Resolution : 1024x768x32 |
In order to compare the nForce2 boards’ performance to that of the KT400, we took the fastest KT400 motherboard currently available, an Asus A7V8X. You will find its benchmark results in each chart.
Benchmark Results
All benchmarks were performed with DDR333 memory at CL2, as this is the fastest configuration for nForce2. However, we also ran our benchmark suite at DDR400. The result is quite surprising, as there definitely are applications that benefit from DDR400.
In addition, we took the fastest board (Asus) and repeated the benchmark suite with only single-channel mode to find out what the real performance gap is between KT400 and nForce2 under the same conditions.
CPU Clock/ FSB Clock
As FSB overclocking within a range of 1-2 MHz can have tremendous impact on benchmark results, we regularly check if the manufacturers adhere to the specifications. Winning reviews by slightly overclocking the system means cheating the customer.
Interesting to see that all three companies stick exactly to the specifications. One reason for this is that the nForce2 chipset does not require an external clock generator.
