[Info-vax] nice for VMS

[David J Dachtera]

Michael Moroney wrote:
> 
> "Fred Zwarts" <F.Zwarts at KVI.nl> writes:
> 
> >>>    What's wrong with "set process/priority=1"?
> >>
> >> Priority zero is better because then it just scavenges CPU time.
> 
> >Are priority 0 processes still competing with the NULL process,
> >or do I carry to much of the VMS history with me?
> >(I know the NULL process is no longer visible in newer VMS versions.)
> 
> I believe the null loop is actually a loop in the executive looking for
> processes to run, so I think it's more than the "NULL process not visible".
> 
> I suppose you could run an experiment:  On an otherwise totally idle
> VMS system (single processor for simplicity), run a CPU-bound process
> at priority 0 for a fixed length of time (say 1 minute) and see how much
> CPU time it uses.  ~1 minute CPU time = no competition.  30 seconds =
> split evenly with "something else".  Might be interesting to do the same
> on an old version of VMS with a visible NULL process to see if they
> really do compete equally.

Funny thing is - on an otherwise idle machine, a CPU intensive process
at DEFPRI will NOT consume 100% of a single CPU. Dunno if that's just a
quirk of the scheduler or what.

D.J.D.