🚀 go-pugleaf

   I've followed the threads of comp.arch for some time.  It may not be the
right place to ask this, but I don't know where else.  If this is out of line,
please forgive me, and point me to the right newsgroup.
   I'm wondering about scheduling of processes in a general purpose interactive
system which has one or more vector processors.  Ideally, the tasks that can
make use of the vector processors get assigned there, and scalar work gets done
on the scalar boards.  But how do you deal with an interactive task (high
priority) which wants to do just a little vector magic on the way?  Do you
allow this sort of overlap, or totally segregate the two?
   Similarly, if there is a backlog of vector jobs to process and the scalar
engines are largely idle, do you schedule some of them to run slowly on the
scalar processors?  Looks like scheduling software for a system like this
could be quite messy.
   Yes, I know this is software.  We have a VAX system (6330) which may become
a 6430.  DEC is rumoured to have a vector board in the wings.  I'm wondering
how well a vector board might be utilized in a general purpose machine, and the
questions above affect that directly.
   If this in fact is the right place for this discussion, I'd like to hear
about how a clever scheduler might try to address these issues.  b.
--
Brent Sterner                        Technical Support Manager, Academic Systems
Network    <BRENT@uwo.ca>            <BRENT@UWOVAX.BITNET>
           <129.100.2.13>            Telephone  (519)661-2151 x6036
Last Gasp  Computing & Communications Services, Natural Sciences Building
           The University of Western Ontario, London, Ontario, Canada  N6A 5B7