lampi@pnet02.gryphon.com (Michael Lampi) (12/28/89)
For all of you confused by the contradictory musings regarding multi-tasking,
multi-processing, etc., here's another term: multi-programming.
Fact: you can not have a multi-processing system without multi-processors;
i.e., multiple processing units; e.g., cpu's. This is regardless of
memory mapping registers, virtual memory, etc.
Fact: the mainframe systems that Mark Crispin speaks of were mostly single
(uni-) processor systems, that executed programs in a
pseudo-simultaneous fashion. This fashion involved, typically, time
slicing (timeshare machines) or priority-based schemes where, typically,
programs ran until they made a system call that required I/O, during
which lower priority programs could be executed.
Multi-tasking is a term normally used to described systems capable of
scheduling for execution more than one task 'at a time', performing context
switching between the tasks as required. This can be performed on a single
processor system or on a multiprocessor system.
Multi-programming is a term normally used to describe mainframe systems where
programs perform voluntary relinquishment of control in order to perform I/O,
transaction processing, etc., and are normally run in a purely priority-based
scheduling environment. Base registers, virtual memory, etc. have nothing to
do with whether a system is multi-tasking, multi-programming or
multi-processing.
Strictly speaking, when your box can do more than one thing at any instant in
time (instant being as short an interval as makes sense), then your box can be
referred to as a multi-processor. The NeXT is not a 'general' multi-processor
since the 56000 is not a general processor as is the 68K; however, it IS a
multiprocessor system.
Michael Lampi MDL Corporation 213/782-7888 fax 213/782-7927
UUCP: {ames!elroy, <routing site>}!gryphon!pnet02!lampi
INET: lampi@pnet02.gryphon.com
"My opinions are that of my corporation!"