Newsgroups: comp.realtime,comp.os.qnx,comp.os.ms-windows.advocacy,comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!EU.net!sun4nl!sci.kun.nl!plm
From: plm@atcmp.nl (Peter Mutsaers)
Subject: Re: Real-time systems: Windows-NT or QNX
In-Reply-To: SCHIEVEL@ix.netcom.com's message of 16 Oct 1994 02:34:48 GMT
Message-ID: <PLM.94Oct17162018@nijmegen3.atcmp.nl>
X-Attribution: PLM
Lines: 24
Sender: news@sci.kun.nl (News owner)
Nntp-Posting-Host: atcmpg.atcmp.kun.nl
Organization: AT Computing
References: <37q3g8$md3@ixnews1.ix.netcom.com>
Date: Mon, 17 Oct 1994 14:21:35 GMT
Xref: glinda.oz.cs.cmu.edu comp.realtime:7187 comp.os.qnx:2240 comp.os.ms-windows.advocacy:40626 comp.robotics:14451

>> On 16 Oct 1994 02:34:48 GMT, SCHIEVEL@ix.netcom.com (Michael
>> Schievelbein) said:

  MS> SAY WHAT ??
  MS> I change priority on my processes all the time,
  MS> ya, it is not as simple as a click of the mouse, but easily done
  MS> with a little rexx file.  The reason they did it that way, was so
  MS> that people didn't go around changing priorities and then complain
  MS> about their performance.

The fact that priorities exist and that you can change them, does not
mean that these are fixed and absolute priorities. Not to mention the
fact that OS/2 (and most other general purpose OSses) is not
preemptive: If an interrupt comes in which makes one of your
tasks/treads runnable again, the current task will first finish its
current time-slice. This is a killer for real-time performance.

For fast real-time systems you'd also better not have the overhead
that virtual memory brings to your context switches...
-- 
Peter Mutsaers                  |  AT Computing bv,  P.O. Box 1428,
plm@atcmp.nl                    |  6501 BK  Nijmegen, The Netherlands
tel. work: +31 (0)80 527248     |
tel. home: +31 (0)3405 71093    |  "... En..., doet ie het al?"
