🚀 go-pugleaf

Fredrik.Pettersson@sto1.wmdata.se writes:

>Dan Pop <danpop@mail.cern.ch> wrote:
>
>[snip]
>
>>Is it?  Pray tell, how is MacOS multitasking the following "application":
>
>>main() { for(;;); }
>
>I did one of those on a NT at work and it really shared processor time nicely,
>100% to that process and nothing to the other applications.

If the other applications needed cpu time but didn't get it, the NT scheduler
is broken.

>So if you write your applications badly they will lock system more or less even
>in a preemtive environment.

Nope.  In a sound preemtive environment cpu-bound applications don't lock
anything.  Try to run N instances of the above "application" in an
otherwise idle VMS or Unix system and you'll see that the cpu resources
are evenly distributed between the N processes:

  PID USERNAME PRI NICE  SIZE   RES STATE   TIME    CPU COMMAND
 1723 danpop    50    0 2312K  147K run     0:01 11.20% forever
 1671 danpop    50    0 2312K  147K run     0:02 11.10% forever
 1752 danpop    50    0 2312K  147K run     0:02 11.10% forever
 1810 danpop    50    0 2312K  147K run     0:02 11.10% forever
 1777 danpop    50    0 2312K  147K run     0:01 11.10% forever
  870 danpop    50    0 2312K  147K run     0:03 11.00% forever
 1668 danpop    50    0 2312K  147K run     0:02 10.90% forever
 1249 danpop    53    0 2312K  147K run     0:04 10.40% forever

The remaining percents were given to the other (non cpu-bound) processes 
running on the system: the X server, the monitoring process which produced
the above output, etc.

cpu-bound applications are a fact of life on real computers, which handle
them without any kind of problems and completely transparent to the
programmer.

>Is it my ignorance or could I be right on this point?

It is your ignorance.

Dan
-- 
Dan Pop
CERN, CN Division
Email: danpop@mail.cern.ch 
Mail:  CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland