🚀 go-pugleaf

MPA15AB!COMPSYSU@TRENGA.tredydev.unisys.com writes:

>The tagged words mean that the A Series archetecture is object oriented,
>with the hardware aware of the basic types of objects.  Thus, the hardware
>prevents improper operations on objects.  The mechanisms used for this are
>quite general.

Grady Booch in "Object-Oriented Designs with Applications" recognises
the A Series as object-oriented on p34. He says "The concept of object
had its beginnings in hardware over twenty years ago, starting with the
invention of descriptor based architectures and, later, capability
based architectures. These architectures represented a break from
the classical von Neumann architectures, and came about through attemtps
to close the gap between the high-level abstractions of programming
languages and the low-level abstractions of the machine itself. According
to its proponents the advantages of such architectures are many: better
error detection, improved execution efficiency, fewer instruction
types, simpler compilation, and reduced storage requirements. Computers
taht have an object-oriented architecure include the Burroughs B5000,
the Plessey 250, and the Cambridge CAP, SWARD, the Intel 432, Caltech's COM,
the ibm system/38, the Rational R1000, and the BiiN 40 and 60."

The Burroughs B5000 was the forerunner of the Unisys A Series. Just 30
years have passed in between. Interestingly, von Neumann got his ideas
from Eckert and Mauchly, who were the founders of Univac, which became
Sperry, which was taken over by Burroughs. The architecture got accredited
to von Neumann, not Eckert and Mauchly, as von Neumann then went to
IBM, with the ideas. In my experience all the advantages that Booch
cites are present in the A Series.

In my last post, I mentioned that A Series architecture makes the
mini/mainframe distinction a nonsense. It also makes a nonsense of
the RISC/CISC distinction. In one way, A Series are very much in the
CISC camp, with string handling instructions, etc. However, they would
be classed as a sensible CISC design. In other ways though, A Series has
some of the properties of RISC design, many of the instructions being
very sensible, and just simply acting upon the two top of stack
registers. Niklaus Wirth designed the lilith processor, which is
a bit like the A Series, but does not have descriptors. He pointed
out that we should have RISC (rational instruction set computers? I think,
but its late on Friday, and my memory has gone if someone can help
out on that one).
-- 
Ian Joyner  ACUS (Australian Centre for Unisys Software)    ian@syacus.acus.oz
"Where is the man with all the great directions?...You can't imagine it,
 how hard it is to grow, Can you imagine the order of the universe?" ABWH
Disclaimer:Opinions and comments are personal.