Possible hardware errors...
 

This started after my PC came up after a Windows update restart 40 mins ago. The error is persistent and has occured 8 times.
Get the message "Possible hardware errors have occured during the test! 1 ROUNDOFF > 0.4"
It's on Worker 6 (thread 11 & 12) on the 3960X. I'm running the CPU on safe, no OC'ing what so ever. Frequency is 3602MHz acording to Aida64. CPU is 43C and Package is 54C. RAM speed is 1333MHz on 1600MHz RAM. And CPU usage is about 50%.
I run latest FW on my RIVE motherboard. And I have only installed the absolute neccessary drivers and programs on a new Windows 7 64 installation.

p95's version?

Windows 64 v. 27..9 build 1

Have you run the torture tests?

It all worked great until the restart I mentioned.
Haven't run the tortures test, but running IntelBurntest 2.54 right now. Made 1'st pass at 83GFlops without faults so far. Running on extreme or whatever the highest is.

Does Prime95 run okay when a booted/restarted in [URL="http://windows.microsoft.com/en-us/windows/start-computer-safe-mode#start-computer-safe-mode=windows-vista"]Windows' "safe mode"[/URL]? :smile:

No. And the IntelBurntest made 5 passes during the night without a single fault. Could something have happend to the datafiles when the computer restarted?

In normal mode, please try Prime95's torture test option for an hour. :smile:

It is possible the data file was corrupted when your computer was shutdown. It is also possible, but much less likely, that the Prime95 program is corrupted. If you had the "2 backups" option checked, there are people here who will guide your number crunching back to health.

The reason we bring up torture tests a lot is because Prime can kill a computer in seconds that would last weeks on another program, and vice versa.

The best example is a person who gets their highly overclocked multi-GPU setup furmark stable for three days and then crashes upon loading up something stupid like Civ 4. I've heard of this kind of thing. The torture tests [I]is[/I] at its core an LL test. It will perfectly mimic the work you are looking to give it.

The message you are seeing states that one hardware error was detected since the start of the test. You may not have an ongoing problem, but rather a rare problem, or even a one-time-only-stray-cosmic-ray problem.

See undoc.txt for options to control how often that message is output (or wait for the LL test to end).

It now crashes at startup when worker #6 starts, sometimes to blue screen.
I'll decrease the CPU speed from 4.4GHz to 3.6GHz again to see how it works.