ken@pdn.UUCP (Ken Auer) (08/19/87)
Does anybody out there have any information about Smalltalk-80 running
on an Atari MEGA ? There was a lot of talk about it after the CeBIT
exhibit in Hannover, West Germany (in February?). It was said to be
developed by the University of Dortmund, and ParcPlace Systems even posted
some fairly impressive benchmarks (considering the price of the MEGA,
rumored to be around $1500 with 4 M running on a 68000) in there
newsletter. I sent of some email to the U of D and received no reply.
Since then, I've heard absolutely nothing. Is someone waiting for the
MEGAs to be "officially available"? Please reply to the world, I'm sure
everyone is interested.
The other question is whether anyone is working on a commercial
workstation based on the COM (Caltech Object Machine) architecture on
which to run Smalltalk-80. There was an article about the machine in
comp.arch in May by Fritz Nordby which really impressed me. Some of
the highlights of the article included:
"Each instruction in the COM is an "abstract instruction": it
includes an opcode and specification of three operands
(typically two inputs and one output), but the operation alone
doesn't solely determine the function to be performed. Instead,
it is the combination of the opcode and the operand types (or,
in Smalltalk-lingo, the receiver and argument classes) which
determine how to execute the instruction..."
"The result is a machine that executes Smalltalk-80 code at
a rate competitive with a Sun 3/160 running compiled C code
(about 0.8*Sun for the sieve of Eratosthenes, which does no
calls; 2.0*Sun for Ackermann's function, which is call/return
intensive)."
"Thus, for a quite similar investment of hardware, the inherent
instruction-level polymorphism of the COM allows each
instruction to do a great deal more work than an instruction on
a traditional machine."
Perhaps Fritz will get psyched and repost the article for us
Smalltalk-types, and give us an update. If the COM is as good as the
article states, developing a workstation using the architecture should
shut up the critics who say "Smalltalk will never make it due to the
inherent performance problems of an interpreted language" or, "in order
to get the performance necessary to make it useful, you need a $40K (or
more) workstation." (I get that last one a lot and, unfortunately, it's
pretty accurate :-( )
Perhaps the hardware has finally caught up with the software, it just
needed some one to figure out how to put it together CORRECTLY...
only 1/2 :-)
--------------------------------------------------------------------------
Ken Auer Paradyne Corporation
{gatech,codas,ucf-cs}!usfvax2!pdn!ken Mail stop LF-207
Phone: (813) 530-8307 P.O. Box 2826
Largo, FL 34649-9981