Fan Noise
As mentioned before, the board uses all passive cooling, so the total sound level is 0 dBA. I also noticed no chirping sounds from lose coils or other components.
Overclocking
When I started tweaking with this board I ran into a hang at POST code C1 (memory initialization) every second reboot. First I thought it's an issue with my BH-5 memory, so I tried other sticks which worked fine. After seeing very inconsistent benchmarks numbers for a few runs I found that the board was often running at CAS3 while it was supposed to run at CAS2, which I set in the BIOS. After more experimenting it became clear that the BIOS seems to have a bug when it comes to storing CAS2 (2.5 and 3 work fine).
- Go into BIOS.
- Set CAS2.
- Save and Quit.
- Reboot.
- Go into BIOS.
- CAS says 3 now.
The reason why the BH5 memory didn't work was because BH5 can not run at CAS3.
What was even more annoying, was that when using BH5 and having the CAS3 bug the BIOS would not POST anymore, not even after clearing the CMOS via jumper. Sometimes I could get it to go through by randomly swapping memory modules and trying to boot with them. Soon the test bench got a huge mess out of memory modules.
Later, out of coincidence I disconnected the CPU fan and the system shut down (Shutdown enable for CPU fan in ABIT EQ) and *poof* now system booted with the cleared CMOS. So every time the system hung at C1 after clearing CMOS I just had to disconnect the CPU fan, let the system shutdown and then everything worked normally.
Now knowing all the issues I set out to work with my BH5 memory: Set CAS2 - don't reboot until done with benchmarks. "Don't reboot" may sound easy but it's a pain for day to day use, and even more annoying when trying to find the maximum FSB which the board can at run without crashing.
With a Swiftech Storm waterblock on the CPU and the multiplier set to 4x we could reach a maximum FSB of 323 MHz. The goal of this test is to find out how fast the board can run without being limited by memory, CPU or cooling.
A maximum FSB of 323 MHz is pretty ok, but I would have expected more from ABIT. However, as long as a board can run over 300 FSB it is going to be fine for 99% of the users.
For a more real-world overclocking score we left the multiplier at 9x and slowly increased the FSB. Since our memory will not be able to run that fast at 1:1, we had to drop the memory ratio to 2:3 which means the memory was running at DDR400 while the CPU ran at 2700 MHz.
Going beyond 2700 MHz at CL2 the system quickly became unstable, at CL2.5 a few extra MHz were possible. Further indicators that memory stability needs to be improved.