Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ihnp4!cuuxb!mmengel
From: mmengel@cuuxb.UUCP
Newsgroups: comp.arch
Subject: Re: Single tasking the wave of the future?
Message-ID: <1423@cuuxb.ATT.COM>
Date: Mon, 7-Dec-87 10:47:11 EST
Article-I.D.: cuuxb.1423
Posted: Mon Dec  7 10:47:11 1987
Date-Received: Sat, 12-Dec-87 16:59:36 EST
References: <201@PT.CS.CMU.EDU> <388@sdcjove.CAM.UNISYS.COM> <988@edge.UUCP> <1227@sugar.UUCP> <151@sdeggo.UUCP>
Reply-To: mmengel@cuuxb.UUCP (Marc W. Mengel)
Organization: AT&T-DSD, System Engineering for Resellers, Lisle IL
Lines: 41

In article <151@sdeggo.UUCP> dave@sdeggo.UUCP (David L. Smith) writes:
$In article <1227@sugar.UUCP>, peter@sugar.UUCP (Peter da Silva) writes:
$ >In article <147@sdeggo.UUCP>, dave@sdeggo.UUCP (David L. Smith) writes:
$ >>I agree with John,  however the ideal towards which we are striving is multi-
$ >>tasking operating systems and single-tasking processors.  Admit it, don't
$ >>you hate having your processor multi-task?
$ > 
$ >Hello, no. Until you have something incredible like 64000 processors in your
$ >box, you're going to find yourself running out of the suckers. If you have
$ >a connection machine on your desk, great! If you've got a transputer, you 
$ >lose.  What happens when you want to start up that 5th or 9th program and 
$ >you've only got 4 or 8 CPUs?
$
$Then one of your processors starts to multi-task.  Of course you never have
$enough processors to handle everything, but it would be nice if you did.
$It's an _ideal_.  
$
$The question was "Don't you hate having your processor multi-task" and the
$answers have been "No, because it has to."  If it didn't have to would you
$want it to?  Timesharing is a necessary evil, and we use it because we have
$finite resources.  


We also use time sharing because even on existing time sharing systems,
there is often a noticable percentage of *idle time*, when every single
process on the system is waiting for i/o to complete.  This is even
true on *IX systems with disk cache hit rates of 80 and 90%.  The truth
is, on systems like sun workstations and the lot, if you take some 
process accounting statistics while you are running your c compile
in background, and rogue in foreground, with emacs running the whole
show, you still have easily 20% idle time on the cpu, even with little 
or no paging activity.

Before you decide that your system isn't as fast as you'd  like it
to be because its multi-tasking, try looking at some system activity
statistics for your system.
-- 
 Marc Mengel	

 attmail!mmengel
 ...!{moss|lll-crg|mtune|ihnp4}!cuuxb!mmengel