Subject: Re: Very Very short completion time???
From: "Stratcat" <none@no.org>
Date: 22/07/2004, 18:03
Newsgroups: alt.sci.seti

"stanmc" <stanmcnOspam@ameritech.net> wrote in message
news:0yNLc.40606$eH1.19208960@newssvr28.news.prodigy.com...
~misfit~ wrote:

Yeah, sorry stanmc, I'm a bit fanatical when it comes to hardware. It's
my
hobby and I get a bit passionate about it. From time to time I have
built
new machines for gaming friends and I wouldn't *dream* of letting a
machine
go out the door that hasn't run Prime error-free for 24 hours on it's
own,
then 24 hours in conjunction with SETI, then checking the MBM5 log for
temps, voltages etc. I also do the same with machines I re-build.


No problem. I knew my unit was performing well, as it didn't exhibit the
kinds of symptoms flakey computers usually do. I just wondered why after
all these years I suddenly run into a unit that completes very quickly.
I won't be running Prime95 anytime soon unless my next computer (AMD64)
acts up. Very boring. They should put some interesting graphics on the
screen. Too bad the guy behind MBM5 is stopping development.

I'm gonna' chirp in here: Put me solidly in the 'Prime it" camp! I overnite
Prime & overnite Memtest my personal builds for 8 - 12 hrs. On
those rare occasions I build a machine for family, I take care to Prime
solidly for at least 12 hrs; but I try not to build machines for family
anymore, 'cuz my 'PC barely literate' relatives often call up all paranoid
over some simple s/w issue that they misunderstood, thinking they have an
expensive h/w issue. This usually occurs around 1:00 A.M on a workday night,
or when I'm trying to do my thing w/my 'ol lady. So no more family builds
for me! If I were to build for a 'paying customer', I would definitely Prime
for 24 hrs, min. Especially if it was for a commercial, financial, or
mission critical application. Then I'd prolly let it go for 48 hrs, min.

As far as Priming a heretofore good running machine, I wouldn't hesitate,
either. There's many ways for a machine to deteriorate over time. OC's can
weaken the cpu over time if you've overvolted Vcore (electromigration).
Heat probs can occur as dust gathers on the HSF (corrosion if copper), or
case openings. Moving the machine or furniture can disrupt airflow, or an
environmental change can occur. The TIM (especially the pads) can
deteriorate, or the HSF/cpu interface can become misalligned during
servicing or moving. And, sometimes, components, such as PSU's, just plain
fail, occasionally in a very slow, gradual manner, where they become
marginal under heavy load, rather than failing catastophically. The myriad
of possible probs occuring over time can become a litany.

So yeah, while it isn't my first diagonistic tool, Prime, IMO, is a very
good standard bearer for me, for testing general sys stability. If it passes
Priming, chances are you've eliminated many possible basic sys issues.

BTW - I'm not just addressing your issue specifically stanmc (IMO you've
gotten a noisy WU, as per my reply, earlier -  though a couple hours
Priming never hurts, IMHO), but rather, stability testing, in general.

My $0.03 (inflation adjusted)
-- Strat