From: Schuster <Schuster AT eev DOT e-technik DOT uni-erlangen DOT de>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Interrupts Maximum Rate
Date: Fri, 11 Apr 1997 14:54:47 -0700
Organization: Lehrstuhl fuer Elektrische Energieversorgung
Lines: 37
Message-ID: <334EB327.3F1F@eev.e-technik.uni-erlangen.de>
References: <Pine DOT SUN DOT 3 DOT 95 DOT 970406144445 DOT 11066A-100000 AT rainbow> <Pine DOT SGI DOT 3 DOT 91 DOT 970407142951 DOT 17332A-100000 AT atmosp DOT physics DOT utoronto DOT ca> <334A5E67 DOT 2E34 AT eev DOT e-technik DOT uni-erlangen DOT de> <Pine DOT SGI DOT 3 DOT 91 DOT 970409112749 DOT 17572A-100000 AT atmosp DOT physics DOT utoronto DOT ca> <334D0181 DOT 3D05 AT eev DOT e-technik DOT uni-erlangen DOT de> <Pine DOT SGI DOT 3 DOT 91 DOT 970410183107 DOT 1012A-100000 AT atmosp DOT physics DOT utoronto DOT ca>
NNTP-Posting-Host: eev6.e-technik.uni-erlangen.de
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp


Peter Berdeklis wrote:
> However, I think the problem addressed by the FAQ is not so much the
> maximum interrupt freq, but how high a freq before you notice a
> significant lag in your (non-trivial) program running in the foreground.
> This is not the case in your experiment I think, or am I wrong?
Yes.
Refering the maximum rate, you're right. But the let's say half of the
maximum 
rate enables pretty much time for calculating, also you can perform DMA
interrupts, 
so that the CPU can work in its cache, or the PCI-grafic card perhaps
does some
transfers,or whatever there is to be done in such a computer.

 
> Of course the rule of thumb offered by the FAQ may vary with program
> type.  A program that does a lot of waiting for user input could probably
> use your timings, but a CPU intensive prog might have noticable lag even
> below 10 kHz.

Ok, i don't want to initiate the holy war of
"Interrupts-cause-to-much-overhead" 
and "Polling-is-something-for-loosers". Just want to give some prooven
numbers.

I prefer the interrupt method, and my systems performs 12 khz wiht 12
channels 
to read, some drawing(with allegro :-), some calculation and I'm pleased
with 
it...


**********************************************
Michael Schuster
E-mail: Schuster AT eev DOT e-technik DOT uni-erlangen DOT de
Lehrstuhl fuer Elektrische Energieversorgung
http://www.eev.e-technik.uni-erlangen.de/