Now to the part you have all been waiting for, benchmarking. SiS is aiming this chipset at the VIA KT400A/KT600 chipsets, at least in their whitepapers. So I decided to pit it against an older KT400 board, the Epox 8K9A2+, which was and still is an incredible board for Athlon XP CPUs.
| Motherboards: | FIC KT-748 |
| Epox 8K9A2+ | |
| Processor: | Athlon XP 2400+ |
| Memory: | 512MB Corsair PC3200 CAS 2.5 |
| Video Card: | FIC ATI Radeon 9700 |
| Operating System: | Windows XP Professional |
| Video card drivers: | ATi Catalyst 4.9 beta drivers |
| Benchmarks: | SiSoft Sandra |
| 3DMark2001 | |
| 3DMark03 | |
| Doom 3 |
SiSoft Sandra – Memory
Using SiSoft Sandra I tested to see how well the motherboard handles memory and the bandwidth you can expect from it. Remember that this is a single channel memory architecture, so there’s no way to use 2 DIMMs for greater speeds. There is a slight difference here with the edge being given to the KT400 board. I was surprised by this, because in SiS’s whitepapers for this chipset they claimed a higher transport bandwidth than the KT400 board, but it just doesn’t translate into performance in this benchmark. The difference is very small, practically negligible.
Futuremark 3DMark2001
Next, lets take a look at the old standby, 3DMark2001. All tests were run at 1024×768 with no anti-aliasing or anisotropic filtering enabled. Here, we have a small difference, with the lead going to the KT400 board. It seems that the lack of AGP 8x being enabled on the FIC board is hampering its performance, as the video card is not getting as much bandwidth to crunch data with.
Futuremark 3DMark03
With 3DMark03 we see about a 500 point difference, or about 12%, which is pretty significant. This seems to be a trend with these two boards. Once again I believe the AGP settings have something to do with this.
Doom 3
Last up in the 3D benchmarks we have Doom 3 from id Software. As I’m sure you all know this game has the most technologically advanced commercial gaming engine around. It uses normal mapping, bump mapping, real-time shadows, the whole shebang. I set the graphics settings to medium detail, which still looks awesome, and ran the benchmark 3 times and took the last score, since Doom 3 likes to warm up apparently. Here we see a much larger difference in the scores, with the FIC board dropping dramatically behind. I really don’t have an explanation for this, but since this game is so graphic-card intensive I would say it’s probably the same thing as the others. I would be interested in seeing scores from someone who could get the AGP 8x to work correctly and analyze the delta.
