IntelBurnTest - New stress-testing program

[Kougar]

I briefly mentioned finding a new stability testing program in my help thread dealing with Vista, figured I oughta share the news here so y'all could have some fun with it.

This program reportedly is much much quicker than Prime 95, reportedly able to detect errors in 10 minutes what Prime can miss in 8 hours of Priming. Now that Rory got my system back into gear I'm going to put these claims to the test myself once I get a little spare time for it.

Link: http://www.xtremesystems.org/forums/showthread.php?t=197835

In the brief time I have tested with it, I can easily say that this program is the absolute best for heating up an Intel CPU. (No idea how well this runs on AMD chips, if at all) Forget CPUBurn and Prime 95 for heating up a CPU, this app even spiked my Q6600 temps despite my WC loop.

Five minutes of Prime 95 using in-place large FFTs maxed out at 49, 49, 45, 46. Just the first run of IntelBurnTest pegged the temps to 57, 57, 55, 55.

 

[Rob Williams]

Posted about this to the front page, thanks a lot for the link. Will definitely use this in the lab from now on. I'm impressed by those temp differences... that's amazing.

 

[Kougar]

Well the app has been updated several more times.

Most notably is how the program runs, is a bit easier to use for a novice user. However more importantly the Linpack libraries were updated, and some how the program is even more stressful than it was previously.

I've had a little time and been playing with it, all I will say is that some overclocks that tested 24 hour Prime stable hard-locked within 3 seconds of testing, and some other overclocks that I knew to be unstable but required 9+ hours of Priming to detect only took a few minutes to find the instability.

Right now I'm at 3.6GHz using 1.5v, temps were at 75c using Real temp. Slowly working my way back down to see what the min vcore is that I can get away with, should be ~1.45v ish...

My problem with Prime was that I never could find a fully stable setting at 3.6GHz, sometimes I could run 24 hour blend or Lage-FFT runs and not find a thing, other times I could run Prime95 and it would error within 30 seconds.