Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!ucbvax!decwrl!hplabs!motsj1!mcdchg!mcdphx!teroach.UUCP!dbk
From: dbk@teroach.UUCP (Dave Kinzer)
Newsgroups: comp.sys.amiga
Subject: Re: AmigaDOS scheduling
Summary: Another viewpoint
Keywords: Compute bound tasks
Message-ID: <10920@mcdphx.phx.mcd.mot.com>
Date: 20 May 89 14:42:50 GMT
References: <216@doctor.Tymnet.COM>
Sender: listen@mcdphx.phx.mcd.mot.com
Reply-To: dbk@teroach.UUCP (Dave Kinzer)
Followup-To: comp.sys.amiga
Distribution: na
Organization: Motorola Microcomputer Division, Tempe, Az.
Lines: 37

In article <216@doctor.Tymnet.COM> jms@doctor.Tymnet.COM (Joe Smith) writes:
[with deletions]
>
>The current priority scheme means that a ray-tracing program at priority -6
>will not get any runtime if there is another compute-bound program running at
>priority -5 or higher.  I would like to see a schedular that would allow me to
>declare that this compute-bound program is to get 1/3 of the background cycles
>and that compute-bound program is to get 2/3 of the background cycles.

   As an example of why this is not as good an idea as it sounds, consider
the following "compute bound" programs.  Program 'A' will require 6 minutes
of CPU time.  Program 'B' will require 4 minutes.  If we launch these with
each 50 percent of the CPU power, Program 'B' will finish in 8 minutes, and
program 'A' will finish in 10 (last 2 minutes at 100% cpu).  For a worst
case, we can pick 60% cpu for 'A' and 40% for 'B' causing both to take the
full 10 minutes.  If we allow only one prioritized task to run, the first
gets done at full CPU power (finishing fast), and the second gets done
no later that it would using shared CPU.  For our example, we launch 'A'
with an Amiga priority of 1, and 'B' with a priority of '2'.  'B' gets
done in 4 minutes (4 minutes faster), and 'A' finishes in 10 minutes (the
same as before.)  The method of time-sharing you propose can only delay
the finish of a program.

   Many managers do not understand this principal either.

   If, however, you want two programs to output data to their user windows
at a proportional rate, that is another story entirely.
  
>Joe Smith (408)922-6220 | SMTP: JMS@F74.TYMNET.COM or jms@tymix.tymnet.com
>McDonnell Douglas FSCO  | UUCP: ...!{ames,pyramid}!oliveb!tymix!tardis!jms
>PO Box 49019, MS-D21    | PDP-10 support: My car's license plate is "POPJ P,"
>San Jose, CA 95161-9019 | narrator.device: "I didn't say that, my Amiga did!"


|    // GOATS - Gladly Offering All Their Support  Dave Kinzer (602)897-3085|
|   //  >> In Hell you need 4Mb to Multitask!  <<  uunet!mcdphx!teroach!dbk |
| \X/   #define policy_maker(name) (name->salary > 3 * dave.salary)         |