home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.programmer
- Path: sparky!uunet!ratsys.com!sunfish.ratsys.com!seant
- From: seant@sunfish.ratsys.com (Sean D. True)
- Subject: Re: realtime application
- Message-ID: <seant.63.0@sunfish.ratsys.com>
- Sender: usenet@uunet!rational (Network News Desk)
- Organization: Rational Systems, Inc.
- References: <1992Sep2.064726.6351@nuscc.nus.sg> <Btyt27.JqD@world.std.com>
- Date: Thu, 3 Sep 1992 16:15:18 GMT
- Lines: 23
-
- In article <Btyt27.JqD@world.std.com> fred@world.std.com (Fred A Putnam) writes:
-
- >If you think you can take a data acq app that is currently running on DOS
- >with a 250 Hz interrupt rate, and get two of those to run in real time in
- >two OS/2 "DOS boxes", you're mistaken. The DOS boxes won't run in real
- >time at all because they are being time-sliced with each other and with the
- >rest of OS/2. The timer ticks are virtualized, and do not track in real time.
-
- >Sorry, but I don't think this plan will fly..
-
- Unfortunately, Fred is clearly right. A more effective solution might
- be to run DOS on a bunch of PC's (even SBCs) and use them to double buffer
- into a network file server. A simple semaphore system using the file system
- could be implemented without much trouble. The data rate into the network
- file server will probably be the limiting step, although you could probably
- saturate your available network bandwidth pretty quickly.
-
- Tools to make this easy should be common. They aren't. If you can settle
- for programs-in-a-can, LabTech Notebook is pretty darn good.
- ================================================================
- Sean D. True
- Rational Systems, Inc.
- seant@ratsys.com -- uunet!rational!seant
-