Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cornell!uw-beaver!uw-june!robertb
From: robertb@cs.washington.edu (Robert Bedichek)
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: Leave the PC on vs. Power it off daily?
Message-ID: <13135@june.cs.washington.edu>
Date: 26 Sep 90 09:32:05 GMT
References: <8axQEga00VY642bGAb@andrew.cmu.edu>
Reply-To: robertb@june.cs.washington.edu (Robert Bedichek)
Organization: University of Washington, Computer Science, Seattle
Lines: 63

In article <8axQEga00VY642bGAb@andrew.cmu.edu> jc58+@andrew.cmu.edu (Johnny J. Chin) writes:
>Power On and Off a PC puts the most wear on your hard drives.  The spin down
>and spin up of the platters causes the spindle to wear out.  Leaving the
>system on all the time allows the spindle to spin with less friction, thus
>less wear.

I've seen this claim before, but never the hard data to back it up.  At
least you have to have some rough equivalence between power cycles and
hours of continuous operation.  What you and others suggest, without
meaning to, I imagine, is that it is better to leave a PC on for a year
than to turn it on for a minute, then off, and leave it for the rest of
the year.  It seems unlikely that this is true.  If it isn't, make it
ten years, or a hundred that the computer is left off after the single
power cycle.

>As for the components of the computer (ie. the silicon chips), the power spikes
>induced by a power on decrease its life (this is more so with the denser VLSI
>chips because of the narrower width of the electrical lines).

What power spikes "induced by a power on"?  I've been looking at power
supply outputs with scopes on and off for twenty years and I've never
seen under or over shoot at power on or off in a commercial supply or
one that I designed.  I'm not saying that it never happens, but you
talk as if it is a fact of life.  And what this has to do with narrower
metal lines I can not imagine.  Do the few milliseconds during which
a power supply rail come up contribute to electromigration?  This seems
far fetched to me.  Do you think it blows holes in the narrower silicon
oxide gates?  I really doubt that.

>The problem with leaving PCs on is the load that is induced on the transformers
>out in the street.  PCs use a switching power supply; what this means is that
>the power supply switches "on and off" 60 cycles per second.  This does NOT
>mean that the PC turns on and off 60 times every second.  Because of this
>"on and off" of the switching power supply, a larger load is demanded from
>the transformers (due to lots of small sudden demands for power).

This makes no sense to me at all.  What is the problem w/ a switching 
supply and power transformers that supply them?  Switching supplies
typically switch at 20 to 30 KHz, not 60 Hz.  Who is this a problem for?
The power company?  The computer user?

>
>Ultimately, what should you do?  Leave it on or turn if off daily?
>Well, my suggestion is ... if you use the computer a lot (ie. more than 10
>hours a day) and use the hard drive a lot during this time (ie. not sitting
>in your word processor all day, but doing reads/writes), I would leave it on.

Intuition, often a fine thing, but what is this intuition based on?

>Otherwise, I'd turn them off at night.  I use my PC about 8-10 hours a day
>with about a lot of reading/writing; I still turn off my PC when I don't plan
>on using it for several hours.  Besides, it saves on the electric bill and
>lowers the chances of getting spikes and brown-outs.
>
>Remember, this is my opinion only.

But so much is stated as if it were fact.

I am not saying that leaving your computer on all the time is a bad
idea.  I am taking issue with the assertions made Mr. Chin in support
of his suggestion.

	Robert Bedichek   robertb@cs.washington.edu