Very early stress test fault
 

I was wondering whether anyone with greater experience of Prime95 and probable causes of errors during torture testing can shed any light on my particular circumstance?

I built a new system last week (specs below) and this week have undertaken the stress-testing process.

What I'm finding is that I get very early (under 1 minute typically) rounding errors.

The system is running at entirely default FSB frequency, CPU frequency/multiplier/voltage settings and Ram frequency/timings/voltage.

CPU and case temperatures are normal whilst running the test (around 38 and 27 degrees Celcius respectively)

I've exchanged the CPU for an entirely new one, tried 4 sticks of RAM, various Bios settings and even powering from different electrical outlets but all to no avail.

One thing I've noticed is that the 12V rail from my PSU is running somewhat lower than desired.
(The other rails are also running slightly low but are within the manufacturer's acceptable parameters).


According to MotherboardMonitor5 I'm averaging the following figures:

Sensor: ASB100 Standard 1:

- 12.00: -11.19V
+12.00: -11.69V
+3.30: +3.24V
- 5.00: - 4.90V
+5.00: +4.87V


Sensor: ASB100 Standard 2

- 12.00: -11.78V
+12.00: -11.78V
+3.30: +3.24V
- 5.00: - 4.90V
+5.00: +4.87V



AIDA32 (Sensor: ASB100 Bach) gives me:

- 12.00: -11.78V
+12.00: -11.78V
+3.30: +3.24V
- 5.00: - 4.90V
+5.00: +4.87V


Bios (hardware monitor) says:

+12.00: -11.77V
+3.30: +3.21V
+5.00: +4.86V


I'm not seeing any voltage spikes upon starting Prime95's torture test and I can't correlate the timing of the rounding errors to anything in particular.

To all intents and purposes the machine is running smoothly although I have spotted the odd unexplained 3DMark2001SE benchmark crash (using default GFX card settings too).


Any ideas would be very much appreciated chaps.

Cheers.


***

AMD Athlon XP2500+
Asus A7N8X Deluxe Rev 2.00
2 x 512MB Twinmos (w/Winbond) PC3200
Thermalright 800U w/ Papst 80mm fan
Antec TruePower 480W
Maxtor 200GB 8MB cache

You could try raising the CPU core voltage by one notch.

After a night of testing it looks as though the culprit may well have been SPD (apparently it's not too great with 2 large stick of RAM).
After configuring the RAM frequency manually I've acheived some level of stability at 199MHz (FSB too).