Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!elroy.jpl.nasa.gov!swrinde!zaphod.mps.ohio-state.edu!rpi!uupsi!sunic!ugle.unit.no!ruve.runit.sintef.no!29195
From: 29195@ruve.runit.sintef.no (Gulbrandsen,Arnt)
Newsgroups: comp.os.msdos.programmer
Subject: Re: Generating delays
Message-ID: <1991Mar11.153409.11426@ugle.unit.no>
Date: 11 Mar 91 15:26:36 GMT
References: <sg2Gwuw?@cs.psu.edu> <1991Mar7.053927.12273@meadow.uucp>
Sender: news@ugle.unit.no
Reply-To: 29195@ruve.runit.sintef.no
Distribution: comp
Organization: University of Trondheim/NTH, Trondheim, Norway
Lines: 14
News-Software: VAX/VMS VNEWS 1.3-4

In article <1991Mar7.053927.12273@meadow.uucp>, py@meadow.uucp (Peter Yeung) writes...
>A more accurate method is to write a counting loop to count how many times
>does it loop when the clock counter advance from x to x+1 and use the
>loop count as a timing base for a timing loop which take the same number
>of cycles to loop once as the initial counting loop. Yes, this implies
>intimate knowledge of the code generated by the compiler or write both
>loops in assembler (can be easily done in TC).
> 
NO! That won't work very well with multitaskers like DESQVIEW, which steal
cycles but leave the system clock intact.

Arnt Gulbrandsen
University of Trondheim, Norway
29195@ruve.runit.sintef.no