Copenhagen what is your setttings?

Hmmm, could it be your ram cannot take it?

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

I am not sure but will bad ram cause Prime95 to fail? Works perfectly at stock speed but even at 101 fsb Prime95 fails (even though everything elese seem to be stable).

<blockquote><font size=1>Svar på:</font><hr><p>I am trying to OC a P4 1800 on a Abit TH7-II with 1GB PC800 samsung ram. Even a slight OC results in a failed Prime95 run. I want to OC but not at expense of slightest reliability.<p><hr></blockquote><p>I simply won't believe that your memory can't cope with a 1% overclock. Think about it; if that was the case, it would mean that your system is running on the edge even in standard setup.

There must be some other logical explanation.

I think we have to turn our attention towards your Vcore. The higher Vcore, the hotter the CPU will get when loaded 100% such as Prime95 does.

Start by setting <b>AGPCLK/CPUCLK(AGP Clock)</b> to <b>Fix</b> and <b>Vcore</b> to <b>1.75V</b>. You will have to use the lastest BIOS, build 38, since there is a problem with the <b>AGPCLK/CPUCLK(AGP Clock)</b> feature in build 37.

Then set the FSB to 101 and run Prime95. Watch the CPU temperature during the test-session. As a rule of thumb the CPU temperature should be kept below 60°C. If it works out well, try increasing the FSB by a small amount. Repeat this until your the system gets unstable. If you suspect a high temperature to be causing the instability you will not be able to increase Vcore more unless you improve your cooling (see <b>Note</b> below). If it's unstable and your temperature is below 60°C, then increase Vcore in steps of 0.05V until it becomes stable again and at the same time keeping an eye on temperature. If you reach a point where Vcore is at 2.20V but your system is still unstable, you have reached the limit and you have to decrease your FSB by 1 until it becomes stable again. You may also try to run the <b>RDRAM Bus Frequency</b> with a multiplier of <b>3</b> instead of <b>4</b> to determine if the memory has reached its limit or if the CPU has reached its limit. If the 3-multiplier works, your memory is setting the limit. You now have to decide whether you want a higher CPU overclock and memory underclock or if it's better to let the memory overclock dictate the limit.

<b>Note:</b>
If you are using Intels stock fan you should be OK. Either your heatsink has been installed incorrectly or your case cooling is insufficient. Case cooling is extremely important with an overcloked P4 Willamette. It developes quite some heat, so a big case with a good vented power-supply and 1-2 extra case exhaust fans is required.

/Copenhagen - P4 Willamette 1700MHz@2109 MHz, Vcore 1.75V@2.20V on Abit TH7II-RAID.

The Abit homepage stinks and is slow like hell. But you can find the BIOS here: <A HREF="http://www.abit.com.tw/eng/download/bios/bios-th7ii.htm" target="_new">http://www.abit.com.tw/eng/download/bios/bios-th7ii.htm</A>

Select the file called TH7H38.EXE

45°C, that's nothing to talk about. Prime95 isn't the most efficient CPU heater, but it's efficient in terms of stability testing. You may also want to try a program like CPUburn to see how high your temperature can climb over a 10 minute period. You can download it from here: <A HREF="http://users.ev1.net/~redelm/cpub4gi.zip" target="_new">http://users.ev1.net/~redelm/cpub4gi.zip</A>


/Copenhagen - P4 Willamette 1700MHz@2109 MHz, Vcore 1.75V@2.20V on Abit TH7II-RAID.

Yes, the links I provided in pc1066 latency show that the more rimms you have the worse you are latency wise.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!

YOu guys are right. If I use two sticks everthing goes fine. I wonder why that is though. I have not tried oc'ing over 105MHz. May be next weekend spend some time with it.