View Full Version : Standard for Stability?
Rainmaker
06-03-01, 05:09 AM
How do you know that a overclocked cpu is stable?
My standard: if a cpu survives repeating loops of 3dmark2000 with prime95 in background.
Prime95 seems to be the most intolerant among cpu stressers so if a cpu survives with this in background, it has to be stable?
wild_andy_c
06-03-01, 07:36 AM
winamp and seti at the same time for a day or so.
asmodean
06-03-01, 01:29 PM
wild_andy_c (Jun 03, 2001 07:36 a.m.):
winamp and seti at the same time for a day or so.
I would've been able to run such stuff at winter/spring with an unstable box. My box, that is. Up that FSB to 143, runs fine, but math gives errors. Not right away, tho.
Prime95 for half an hour, it'll error if it's not stable. PiFast3.3 for 4 runs before that, so if it's not quite stable, it'll bum out in them.
outhouse
06-03-01, 01:43 PM
The best standard would be the one where your machine has no lock ups at all, and no BSOD and no random reboots while gaming heavily, Seti, Prime, SiSandra or any other benchmark. I get one error of any kind and i lower my settings until i figure out how to get around said stability issues. it may be a matter of personal choice some can put up with a occasional error but to me thats not stable.
William
06-03-01, 02:49 PM
hmm, when going from a 66mhz bus to an 83 mhz bus your computer becomes more stable.............
Seriously, you can run the tests and all, but only doing what you always do will really tell you. My computer never crashed, well except for explorer, but that was after upgrading to 5.0. Prime95 is good, winamp and Folding are good too. Winamp, folding, talking on the internet, overclockers.com, and downloading stuff is good too!
Although madonion and seti may be good markers, I found running distributed.net (http://http.distributed.net/pub/dcti/current-client/dnetc-win32-x86.zip) to burn in and subsequently test the cpu for 24 hours is pretty safe (at least its been consistent for me).
I'm not sure that running these other applications places any special burden that distributed.net's software doesn't already place on the cpu.
asmodean
06-04-01, 02:21 PM
If the program that is used for math stability testing doesn't check for computational errors, it's useless.
OpenFriday
06-04-01, 03:23 PM
Ive been running Folding and 3dmark 2001, if it holds up to that it's stable in my book.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.