Searching for Mersenne Primes (prime95) and using a OCed system.

[ty1er]

To suck up the spare CPU cycles, i run Prime95 to search for Mersenne Primes on my PC.
Lately i have been testing out different Cpu overclocks with my AMD64 rig. To test the stablity i will run prime95 Large or Small FFTs test. When Testing my system i Do Not run PRime95 to search for Mersennes, I run it in Torcher Test mode.
With my system running around ~2300mhz i seem to be getting errors around the 4 hours to 7 hour Mark.

So my question is: Say if i left my system at this 2300 speed, would there be any dangers/Miscalculationsis running prime95. OR is there a double check or something when just running Prime95 when searching for prime? If there is an error will the program know it? and will it try again to test for the correct sum?

 

[qbek]

If Prime95 reports an error, chances are (IIRC well over 50%) that the whole calculation is wrong and the test will need to be repeated by someone else. Actually, it might even slow down the progress of the project as a whole, because if the residues of the second test will not much with yours the third test will be necessary.

Therefore, as long as your OCed system produces errors chances are that you will not contribute to the project in any positive way...

 

[ty1er]

Originally posted by: qbek
If Prime95 reports an error, chances are (IIRC well over 50%) that the whole calculation is wrong and the test will need to be repeated by someone else. Actually, it might even slow down the progress of the project as a whole, because if the residues of the second test will not much with yours the third test will be necessary.

Therefore, as long as your OCed system produces errors chances are that you will not contribute to the project in any positive way...

yeah, that was i was kinda thinking.....

i'll have to do extensive testing with prime95 Large/Small FFTs (12 to 24 hhours.) before running prime95 to search for mersennes...

 

[FrustratedUser]

Originally posted by: ty1er

Originally posted by: qbek
If Prime95 reports an error, chances are (IIRC well over 50%) that the whole calculation is wrong and the test will need to be repeated by someone else. Actually, it might even slow down the progress of the project as a whole, because if the residues of the second test will not much with yours the third test will be necessary.

Therefore, as long as your OCed system produces errors chances are that you will not contribute to the project in any positive way...

yeah, that was i was kinda thinking.....

i'll have to do extensive testing with prime95 Large/Small FFTs (12 to 24 hhours.) before running prime95 to search for mersennes...

Just don't OC while crunching numbers.