Prime95 fails only when testing sensors with RealTemp


May 11, 2009
First, my system specs:

Asus P5Q MB
Intel E5200 overclocked to 3.33Ghz (266 x 12.5)
4GB Corsair 800Mhz memory (4 x 1024MB)
XFX GTS 250 VC (stock settings)
Noctua NH-U12P CPU cooler
*Only changes made to overclock system were the FSB and VCore, all other selections are manually set at stock levels.

Now my problem:

I have successfully run Prime95 small FFT's for 20 hours without any errors. I have always had to manually shut Prime95 down. However, I just downloaded realtemp (been using SpeedFan before) and whenever I do the sensor test, one core fails usually around 60%-70% test. My thought is that RealTemp drops the voltage to the cores to reduce the percentage load placed on the system and that this is causing Prime95 to fail. Is this right? Thanks for all your help.


Real Temp enables clock modulation during this test, which to my understanding means that the CPU is forced to "sleep" every x number of clock cycles, depending on what modulation setting is chosen. It doesn't change voltages as far as I know.


Sep 11, 2007
Old thread but just to clear things up, RealTemp does not adjust your core voltage during its CPU Cool Down Test. Randomizer is correct. It adjusts the Clock Modulation feature of the Intel processor. Do a Google to find out more about what this does.

I've run this test many times without ever having a problem or my computer freezing up. If your computer is not stable during this test then it is likely that your overclocked CPU is not 100% stable.