Message-ID:	<3D543EAE.7AC758AD@yahoo.com>
From:	CBFalconer <cbfalconer AT yahoo DOT com>
Organization:	Ched Research
X-Mailer:	Mozilla 4.75 [en] (Win98; U)
X-Accept-Language:	en
MIME-Version:	1.0
Newsgroups:	comp.realtime,comp.os.msdos.programmer,comp.os.linux.development.system,comp.os.linux.embedded,comp.os.msdos.djgpp
Subject:	Re: OS, tools selection
References:	<797576f5 DOT 0208091319 DOT 19be6ab6 AT posting DOT google DOT com>
Lines:	42
Date:	Fri, 09 Aug 2002 22:19:08 GMT
NNTP-Posting-Host:	12.90.168.147
X-Complaints-To:	abuse AT worldnet DOT att DOT net
X-Trace:	bgtnsc05-news.ops.worldnet.att.net 1028931548 12.90.168.147 (Fri, 09 Aug 2002 22:19:08 GMT)
NNTP-Posting-Date:	Fri, 09 Aug 2002 22:19:08 GMT
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

Kel Tyree wrote:
> 
... snip ...
> 
> I am about to dive into a semi-embedded project (it would be nice
> if the system were available as a general purpose PC, though that is
> not absolutely necessary) that is focused on a legacy *ISA* data
> acquisition card.
> 
> The interface to the board is fairly straightforward, with a few
> mapped I/O ports and a single HW interrupt line.  The stickiest part
> is that about 1500 bytes must be transferred (to RAM from a fixed I/O
> address) at each interrupt (they're about 200 microseconds apart).
... snip ...

I suspect you are going to have trouble.  Start with the max speed
of 8 MHz, and assume a 16 bit wide system (limits of the ISA
card).  Transfer time will take a minimum of 750 * 125 nS, or
about 100 uS, IF you can find a DMA mode to do that.  It will
probably require at least twice that, and there goes your time
budget.  Nothing is left to handle the interrupt proper, nor to do
any work between interrupts.

If you move data over the IDE buss you have a much better chance. 
You are talking about a data rate of 1500 bytes per 200 uS, or
about 7.5 Mbytes/sec.  To do any significant work outside the data
transfer the burst rate must be considerably faster, so I would
start by looking for a path with a burst rate in the 16 to 20
Mbytes/sec.

Interrupts can be disabled for some time in the DJGPP system,
which I assume you are thinking of using since you posted here. 
It is NOT a real-time system.

Look to the fundamentals first.  There will be enough other things
to byte you anyhow.

-- 
Chuck F (cbfalconer AT yahoo DOT com) (cbfalconer AT worldnet DOT att DOT net)
   Available for consulting/temporary embedded and systems.
   <http://cbfalconer.home.att.net>  USE worldnet address!

- Raw text -