Hardware Tests
I also tried the 153 MHz FSB setting, using an unlocked Pentium III processor. As expected, there was no real success running the main memory in CL2 mode: The system crashed from time to time. After changing the settings to CL3, everything became more reliable. However, I faced several lock ups which I could not definitely reproduce. After switching down to 148 MHz FSB, the lock-ups disappeared completely.
In the next test I mixed three different SDRAM modules. One Crucial/Micron DIMM, one from Memory Solution (Germany) and a third from Viking Components. All three comply to the PC133 standard, but only the Crucial memory is capable of running at a CAS latency time of two cycles. To get the system running stable with all three DIMMs, I had to reduce the RAS-to-CAS delay to 3 clocks as well. After that, the system ran flawlessly.
Finally, I equipped the test system with some additional hardware to stress it a little bit: One older Adaptec 2940UW, a 3COM 905TX PCI network card, a Creative Sound Blaster 128 PCI, a Hauppauge WinTV/Radio card and an AVM ISDN Fritz! Card. Each requires an interrupt, which can give the motherboard BIOS quite a hard time.
Luckily, everything worked at once. That's something that I really appreciate, since setting up a computer can easily become a pretty hairy affair if you have to find IRQ configuration mistakes caused by the BIOS. The worst scenario is a fully equipped motherboard that shares interrupts between several PCI devices without giving you the option to influence it. Having only one card that requires an unshared IRQ (e.g older Promise IDE controllers) will ensure that your whole system will hardly ever work properly.
