Re: An important fact about real time computing

Phil Wilshire (philw@zentropix.com)
Thu, 20 Jul 2000 04:21:21 -0700

yodaiken@fsmlabs.com wrote:
>
> On Wed, Jul 19, 2000 at 10:36:00AM -0600, Jesse C Cronce wrote:
> > task (as per your example) that's one thing, but I thought that RT
> > scheduling
> > is about deadlines and whether or not they can be met, which has nothing
> > to
> > do with fast computing.
>
> That's true, but it's also true that the shorter the latencies the
> greater the range of applications. So, HP-UX "RT" offers determinism
> up to some millisecond point and RTLinux offers determinism at the
> 15 microsecond level -- one can be used for robotics and one cannot.
>
> victor
>

It really depends on the speed of the robot perhaps ???

My take on all this is that Real Time is really all about worst case
tolerances on schedule or interrupt deadlines.
Speed is an issue when the tolerances get tight.

If you can stand 100's of milli Seconds then you will probably be OK
most of the time.
Once you start getting below 5 milliSeconds then you need to listen in
to the Real Time mailing lists.
Get below 1-2 milliSeconds then you can go NOWHERE else but Real Time.

IMHO it is unrealistic to attempt to use regular linux without a Real
Time Extension ( RTAI or RTL ) in this range.

You may get it to work on one system / configuration but as soon as you
change something or add another driver you stand the chance of breaking
the application.

Regards
Phil Wilshire

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/