....acquainted with dividers
Okay, so when you're safely into Windows once again, run the Sandra memory bandwidth benchmark and again note your results.
Run
Orthos/Prime/SP2004 again. Our preference is for the Blend test for all
these checks, so that everything gets a good stressing.
Remember this is just an exercise to get you acquainted with dividers.
Have a mess around with some of your regular programs again, see if you can notice any difference.
Of
course, it doesn't end there. Those of you with faster RAM can go back
to the BIOS and choose a higher divider and see just how high you can
go.
But what we're going to look at next is trying to find a
good balance between a high CPU overclock, a high RAM divider and the
tightest timings possible.
At this point, we're assuming
that you've tested how high your RAM will go and have made a note of
the speed. And the same for the timings and for the CPU overclock -
which you might have saved to disc if your motherboard allows. If so,
load up your best stable CPU overclock.
Now choose a divider that is as close to your RAM's max as possible but without going over.
Keep the RAM timings standard for now.
Set the RAM voltage to whatever you had it at when you found it's max speed.
To
aid stability, you can also change the northbridge setting from AUTO to
something like 1.55V. This isn't absolutely necessary but you can
always change it back later.
Save, exit and boot to Windows.
This time, if the PC doesn't boot or if it hangs or gives you a BSOD while loading Windows, there are a few things you can try;
A different divider, usually (and preferably) lower but sometimes a higher one will work.
Or a notch more voltage to the northbridge.
Or a slightly slower FSB, try dropping it 10MHz at a time.
Once
things are stable, you can try the tighter timings but you may need to
drop the FSB by between 30MHz and 50MHz then work your way back up.
All of this is just a balancing act and after a while you'll get a feel for what will and what won’t work.
Don't worry if something doesn't work, just keep experimenting - you'll get there eventually.
