I've found that running Memtest-86 (
http://www.memtest86.com) is a pretty fast way of
evaluating whether you've gone too far on the memory overclocking. My best test for CPU
OC'ing stability so far is MPEG encoding (crashes real fast if I've gone too far!).
Until I bought this motherboard, my mind was still stuck thinking SDRAM thoughts: I thought
that the faster the memory was running, the better the overall performance would be.
That doesn't appear to be an absolute. I started out with two 256MB Spectek (Micron) DDR333
sticks (the Albatron dealer didn't have any more DDR400 memory).
I found that the best performance with my 2.4C was to use the CPU/DDR ratio setting
of X1.33 (also called 3:2) and never let the memory get even near it's rated clock speed.
My best Sandra memory test score was 4867 & 4854 for the DDR333 memory. Of course, this
memory hardly stands up to comparison with the memory your system has (but hey... I'm cheap!)
A similar experience was written up in an article on
www.overclockers.com by Ed Stroligo on 6/6/03
on his experience with his Abit IC-7 , with a 2.4C and two sticks of Corsair XMS3500. He also concluded that max'ing out the FSB while undeclocking the memory gave him better overall performance. I guess I'm not a real overclocker: I can't justify buying DDR466 or DDR500 memory
pairs for 200 or 300 MHz of possible speed increase.
I am running my CPU and memory at stock voltage: I proved very fast with Memtest that
overvolting the Spectek memory wasn't helping me at all. Ditto with overvolting the CPU as well.
Right now my CPU is running 38 deg C. with a stock Intel heatsink. It gets up to about 43 deg
after running all day.
I'm looking REALLY HARD for something to complain about with this motherboard, but I'm still
happy with the hardware. I'm even beginning to not complain about XP Pro as much: hyperthreading
processor support is worth it!
happy at getting 3.3 GHz out of a 2.4 GHz-rated (and priced) CPU.
much happier with it than my previous MB, an Abit BX6 Rev2.