A guide to stability testing

Just been linked to this. Nice guide. Cheers.

One thing about the Super pie settings: The K amd M in the options mean thousand and Million digits not bytes as you've listed them as Kb and Mb.

Cheers.
 
w3bbo said:
SP2004

Very similar to Prime 95 with a neater GUI interface. There are reports however that Prime95 will throw up errors that SP2004 dosn't catch (even though they are in effect the same program.
Click to expand...

That's because SP2004 is basically an old unoptimised version of Prime95, using old code which doesn't push current processors as hard as the latest Prime95 version now does.
 
Shrapnel said:
Just been linked to this. Nice guide. Cheers.

One thing about the Super pie settings: The K amd M in the options mean thousand and Million digits not bytes as you've listed them as Kb and Mb.

Cheers.
Click to expand...

Thx for the correction - edited guide.

As for SP2004 there is actually a new version out (has been for a while now afaik) although it was in beta form. The new one is based on P95 24.14 although having tested it, I personally have still found P95 to be more sensitive to errors even though they apperently use the same verson of Prime. I'll update and modify the guide later today with pics etc.
 
.::lawrywild::. said:
also, Rosetta@home for a few days constant is a good long term stress tester...
Click to expand...

NO!

Running Rosetta or any other distributed computing project as a stess test is a bad idea because an unstable system can give incorrect results. You don't want to be the guy who missed the cure for cancer just because his rig was unstable, do you?
 
Updated the guide :

Added new beta version of SP2004. This version is based on the Prime95 source ver.24.14. It will also give you readouts of CPU and PWMIC temp, voltages and fan speed so long as you have motherboard monitor 5 installed and setup correctly. As this is still in 'beta' form I am reluctant to make the switch to sp2004 from P95 at this time.

Also worthy of note is that for some reason SP2004 pages the hard disk when running the 'blend' test which might deteriorate the HD if used for extensive periods, especially if the HD dosn't have adequate cooling as this stresses the HD and so data loss could occur as a result. Obviously I can't say for sure that this WOULD happen but bettter to be safe than sorry imo. I therefore wouldn't recommend running the blend test if you are after testing the memory and should stick to using memtest86+,SuperPI or large fft's with either P95 or SP2004.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom