System setup and benchmarking notes
Here's a quick rundown of the test system should you wish to compare benchmark results with your own.
- Intel Pentium4 1600MHz S478 Northwood CPU Northwood at 2133MHz / 133FSB
- Shuttle SiS648 AS45GTR motherboard in DDR-333 mode.
- ABIT SR7-8X SiS648 motherboard in DDR-333 mode.
- ABIT IT7-MAX I845E run in DDR354 mode
Common components
- ASUS GeForce4 Ti 4400 128MB at stock speeds (275/550)
- 256MB Corsair XMS3200 C2 run at Ultra timings in DDR333 mode
- 120GB Western Digital 120JB 7200rpm hard drive with 8MB cache.
- Liteon 16x DVD
- Samcheer 420w PSU
- 21" Sony G500 FD monitor
- Thermaltake S478 cooler
Software
- Windows XP Professional Build 2600.xpclient.010817-1148
- SiS v1.10.03 AGP and IDE drivers
- Detonator XP 29.42 drivers
- Sisoft Sandra 2002 Professional
- Pifast v41
- Lame v3.91 MP3 encoding with Razor-Lame 1.15 front-end
- Virtual Dub DVD encoding, DivX 4.12 CODEC
- 3DMark 2001SE
- Comanche 4 benchmark
- Serious Sam 2 Demo
- Quake 3 v1.30
I'll be using a 1.6GHz Northwood A running at 2133MHz / 133FSB for these tests. This processor is also useful as it has seen 165FSB in other motherboards. This should, in theory, help us find out just how far we can push the Shuttle AS45GTR.
First thing to note was that the motherboard ran flawlessly at its specified 133 CPU FSB and 166MHz memory speeds. I encountered no crashes, anomalies, or discrepancies whatsoever over 2 days' worth of use, even when I used the strictest timings possible.
The situation, however, was markedly different when trying to run at the unsupported but implemented DDR-400 specification. I used a proven module of perhaps the fastest memory currently available in the form of 256MB Corsair XMS3200 C2. This module has seen DDR-450 speeds in an ABIT IT7-MAX motherboard. Even raising the DDR voltage to 2.75v, the maximum permitted amount, didn't solve my stability problems. I could complete my benchmarks at DDR-400 if I lowered the timings to amongst the slowest possible. Even then, I wouldn't have called it supremely stable. I also tried some DDR-400-capable Samsung PC2700 memory and had similar problems.
With that in mind, I will not include benchmarks at DDR-400, solely because I don't feel happy with them. This is not an isolated incident, though. I had similar trouble running DDR-400 with the SiS648-equipped ABIT SR7-8X, although that was more stable. Changing DIMM slots made no difference whatsoever.
Overclocking via manipulating the FSB proved to be something of a disappointment. I'm sure this has more to do with the chipset than this particular motherboard. I managed to gain semi-stability at 150FSB with a processor that I know can hit 165FSB in i845E motherboards. This seems to be a trait with SiS648 motherboards, perhaps actively cooling the Northbridge may help ?. The fact that you can't lock the PCI and AGP busses, a la ABIT, doesn't really help either.
