Real-Time Scheduling and Ada

Real-Time Scheduling and Ada

[Andre N. Fredette]

The primary real-time scheduling algorithms for uni-processors seem to
be rate-monotonic, earliest deadline, and least slack.

The general Ada runtime system seems to be fairly compatible with
rate-monotonic scheduling.  Furthermore, with the implementation of some
of the features outlined in the Catalogue of Interface Features and Options
for the Ada Runtime Environment (CIFO) and Ada 9x documents, such as
priority inheritance and different queuing disciplines, Ada and
rate-monotonic scheduling should be very compatible.

However, Ada runtime systems do not seem to support earliest deadline or
least slack scheduling at all.  As far as I can tell, there is no explicit 
support for these disciplines in Ada 9x either -- although their
implementation in a particular compiler is not prohibited.

I'd be interested in any comments on any of these ideas/statements.

Andre N. Fredette
anf@pesun.ncsu.edu

Re: Real-Time Scheduling and Ada

[Thomas Ralya]

In article <1991Nov19.214910.23519@ncsu.edu>, anf@pesun.ncsu.edu (Andre N. Fred
ette) writes:
|> 
|> The primary real-time scheduling algorithms for uni-processors seem to
|> be rate-monotonic, earliest deadline, and least slack.
|> 
|> .
|> .
|> .
|> 
|> However, Ada runtime systems do not seem to support earliest deadline or
|> least slack scheduling at all.  As far as I can tell, there is no explicit 
|> support for these disciplines in Ada 9x either -- although their
|> implementation in a particular compiler is not prohibited.
|> 
|> I'd be interested in any comments on any of these ideas/statements.

I'd like to make two points that may be a paritial answer:

 1) rate monotonic scheduling and the Ada runtime model are based on fixed
    priority scheduling, while earliest deadline and least slack are viewed 
    as requiring dynamic priorities (leading to higher executive overhead).

 2) in 25 years of building real time systems, I've never heard of a real
    time system using either of the dynamic priority scheduling techniques.
    (I'm sure they're out there somewhere, but I wonder if they are in
    wide-spread use.)

Re: Real-Time Scheduling and Ada

[Dale Lancaster]

In <35430@as0c.sei.cmu.edu> tar@sei.cmu.edu (Thomas Ralya) writes:


> 2) in 25 years of building real time systems, I've never heard of a real
>    time system using either of the dynamic priority scheduling techniques.
>    (I'm sure they're out there somewhere, but I wonder if they are in
>    wide-spread use.)

I know of one place that used it for handling multiple realtime flight
simulations at the same time on the same computer (earliest deadline).
However you could not have an abritary number of processes, you had
to carefully make sure there were enough time slices to service all
compute needs for all processes running.  It seems to have worked well,
but they are now moving towards multi-processors with one process per
processor.

dale
--
===============================================================================
Dale M. Lancaster			    Work: 214-497-4581        
Convex Computer Corporation		     Fax: 214-497-4848        
Product Manager				Internet: dale@convex.com 

Re: Real-Time Scheduling and Ada

[Kurt Fuchs]

In the article from 19 Nov 91 21:49:10 GMT
anf@pesun.ncsu.edu (Andre N. Fredette) writes:

> However, Ada runtime systems do not seem to support earliest deadline or
> least slack scheduling at all.  As far as I can tell, there is no explicit 
> support for these disciplines in Ada 9x either -- although their
> implementation in a particular compiler is not prohibited.

I once browsed through the ADA 9X working group proposal. From that I
can verify that no functionalities, which would handle deadlines or
combinations of deadlines + value of the task to the system
(time/value function scheduler) will be incorporated in new ADA.

rgds
--
--

Kurt Fuchs				  voice:    +43 (1) 39 16 21 / 177
Alcatel Austria-ELIN Research Center      fax:      +43 (1) 39 14 52
Ruthnergasse 1-7			  Internet: fs_fuchs@rcvie.at
A-1210 Wien				  UUCP: ...mcsun!tuvie!rcvie!fs_fuchs
Austria/Europe

-- 
--

Kurt Fuchs				  voice:    +43 (1) 39 16 21 / 177
Alcatel Austria-ELIN Research Center      fax:      +43 (1) 39 14 52
Ruthnergasse 1-7			  Internet: fs_fuchs@rcvie.at
A-1210 Wien				  UUCP: ...mcsun!tuvie!rcvie!fs_fuchs