In September 2000, I purchased a Duron 600 and MSI-6330 with the key aim of
overclocking it. At that time the MSI-6330 motherboard didn't really have that
much going for it as an overclocking board. Sure you could up the CPU core
voltage by 0.05Volts and the FSB was fully adjustable in 1MHz steps, but it
had no multiplier adjustment (this is the version 1 board).
However, I did find this article
describing how to modify the MSI 6330 board so that you could adjust the
multiplier. The mod was relatively simple and required only cutting two tracks
on the motherboard and then running several wires to a small board with
a number of jumpers on it. Not only could the board change the multiplier,
but you could substantially increase the core voltage too.
With this board and the original 'generic' heatsink that came with the Duron
600, I was lucky to get it to run stably at 700MHz. So I went out and bought a
'Chrome Orb' fan to cool the CPU down a bit more. This worked wonders and
I could stably run at 800MHz. To achieve this I think the Vcore was set as high
as I could; about 1.8V.
But I wondered if I could go faster. I found another hardware mod on the net
(now disappeared) describing how to wire a potentiometer onto your voltage
regulator chip to get even higher Vcore voltages. I did this, but deliberately
put a resistor in series with the Pot to limit how far I could go. It would
allow for a max Vcore of 2.02Volts.
Further testing ensued and I could run at 850MHz now, but for some odd reason
my CPU would never even turn on with a 9x or 9.5x multiplier. So what I ended
up doing was to run the FSB at 105MHz with an 8.5x multiplier to 893MHz...
Reasonably close to 900MHz. I can't quite remember what voltage this required,
but probably about 1.95Volts.
This 893 setup ran perfectly for several months. I then started playing with
higher FSB frequencies; 6 x 120 (ie 720) would work. But then I tried some
other combination and the machine would not boot.
I originally thought this could be deteriorating pencil marks, so out came the
motherboard and I redid the pencil marks. No go. Still dead. The MSI board has
some lovely LEDs that tell you how far its gotten through the POST sequence.
These were all red meaning it hadn't gotten anywhere. I reset the CMOS, No go.
Tried the default voltage and default multiplier. Erased the pencil marks
entirely. No go.
Either my motherboard or CPU was dead. Given that I had modified my
motherboard, I spent some time going over the connections to my addon board.
Nothing looked too out of place.
In the end I had a gamble and bought a new CPU and fan; a Duron 750. Put it in and
it all worked again. I had somehow fried my Duron 600. I'm still not sure why.
Some possibilities are:
- Electron migration
- By running the FSB so high for a short period, the processor overheated??
- The heatsink compound wasn't up to scratch. Since the heatsink had been
previously removed and put back on several times (those bloddy pencil marks),
there wasn't a good thermal path.
So admitedly I was a bit hesitant about overclocking this new Duron ... but of
course I got curious. Here are my results for stable running:
The magical 1000MHz would appear to require more than 1.9Volts
and I'm hesitant about doing this as I'm wondering whether my PSU is up to
- 9x 100 @ 1.7V works
- 9.5x 100 @ 1.85V works