IntelBurnTest - The new stress-testing program

[rge]

I also am BurnTest stable times 20 iterations but fail out in Prime95 in less than a minute. Very modest overclock from default 3.2 GHz to 3.4 GHz. I am running 64-bit Vista. Not sure what's up here. Voltages are the same running both BurnTest and Prime95.

I had same problem with original linpack in previous thread with orthos/prime showing error long before linpack did.

However, using newer libraries here and probably more importantly using setting 1 which forces max memory usage, I found the opposite that intelburntest is more efficient than orthos/prime.

But anything that prevents max memory usage, at least on my setup, makes intelburntest worthless, including using any setting but 1.

Pic is with vcore set to 4 notches below stable, where orthos 10ffts, prime small usually errors out by an hour or less. Linpack setting 3 passing 10 passes. Linpack setting 1 would not even pass two consective passes...fails every pass, output is like a random number generator. prime small ffts 15 mins and still going.

Would be interesting to see pics of what memory usage you are using...whether it is that or just very different for different components.