Good enough stress test to say it's 100% stable?

  • Thread starter Thread starter str
  • Start date Start date

str

str

str

Soldato
Joined
18 Oct 2002
Posts
3,052
After lots of testing with various settings to find the max overclock I've finally decided on running my Sempron 2800+ at 2664 MHz with Vcore at 1.65v (it drops ~0.05v lower at idle and increases ~0.03v higher on load).

stresstestat9hrs4wf.png


The above is 9 hours into a 24 hour Prime95 Blend test and I'm wondering is it enough of a stress test with Prime95 priority at 7? I'm also using the system to download torrents, surf the 'net, and occasionally I watch a downloaded TV episode or movie using VLC.

Or should I not use the system at all and run Prime95 at priority 10 (it's highest) and just leave it doing that for 24 hours for it to be considered 100% stable?

I'm doing the Prime95 Blend test as that's the one that failed the most during overclocking testing but should I at least do the Small FFTs for 24 hours as well or maybe the In-line FFTs is better or even do both?
 
Yeah so it's like no matter how much stress testing occurs it could still fall over sometime in the future. In general though I've read 24 hours is considered adequate for most overclockers for something to be considered Prime95 stable?
 
You should try a variety of tests, a lot are much quicker then prime. I think a 24hr prime is 100% stable but then you could test it in winter and have problems in summer still

OCCT & S+M

Occt is only 30 mins long
 
Haven't used OCCT before but will give it a try soon. I like S&M as in previous testing I found it increases the CPU temp an extra 8c compared to Prime95 so should be a good test to try again after I'm finished with Prime95.
 
Just some input, a lot of people I know on MSN have been experiencing power fluctuations over the past few days, mainly in the afternoon and evening.

They are bad enough to cause my UPS to kick in, and my stand fan to noticably slow down. The dips can last up to a few seconds. I've not noticed anything odd on my PC or my server, but a PC I have downstairs (old one, cheap PSU) crashes badly every time it happens!

I'm in SW Wales, lots of people that I know are experiencing it are in England, so it must be pretty widespread.

Just a thought, as these dips could potentially cause your primes to fail.
 
I'm in Edinburgh and the power went out the other day for about 30 seconds. Was a big shock when the system powered down especially as it was running a stress test at the time and at first I thought it had spectacularly failed. :)

But yeah that's a good point about power fluctuations causing stability testing to fail.
 
NightmareXX said:
Whats snm?
Click to expand...
S&M/S+M/SnM (all same program) is a CPU burn style program. It's good but it's also considered reasonably dangerous to your hardware if you have bad cooling or whatever as it's real good at heating up the CPU. I guess it's advisable to be confident about your setup especially if you intend to leave it running unattended.

You can download it from:

http://www.benchmarkhq.ru/fclick/fclick.php?fid=301

The homepage is here but is in russian. :)
 
Last edited:
My system survived the 24 hour Prime95 Blend test:

24hrprimeblendstable1wh.png


:D

Gonna do some S&M testing now to see how hot it gets compared to the above.
 
You must log in or register to reply here.
Back
Top Bottom