MasPar, anyone?

From: Roger Ivie <IVIE_at_cc.usu.edu>
Date: Wed Sep 11 01:18:06 2002

Qstieee_at_aol.com said:
> I believe it stood for Massachusetts Parallel processing, or perhaps
> Massively Parallel processing. It was for real-time high performance use.

MasPar was a company out of Milpitas, CA founded (I understand) by
ex-DECcies. They built a SIMD massively parallel machine. The unit
was controlled by a front-end computer, originally a VAXstation 3520
running Ultrix, later some flavor of DECstation 5000 also running
Ultrix.

They used the VAXstation 3520 because they got a real good deal on them
from DEC, who was anxious to sell the thing. I designed the interface
between the 3520 and the MasPar which, AFAIK, was the only 3rd party
interface done for the 3520's proprietary MBUS.

They claimed to have licked the problem of compiling for and debugging
SIMD machines using standard languages; IIRC they supported FORTRAN and
C. Seems to me they were more proud of their debugger than their
compilers.

I debugged the front-end interface hardware by poking hand-assembled
code into the console using CP/M KERMIT's TRANSMIT command on my (then)
trusty Kaypro 10. When I arrived with the prototype and my Kaypro, they
said "How quaint; he hand-assembles code". After a while it was "Wow!
Scope loops you can see on a scope!" (they were using a high-level
interpreted thingy called hdb for hardware debugging; their scope loops
were too slow to see on a scope) By the time I left, it was "show us
how you did that." Sometimes being an old fart pays off.
-- 
Roger Ivie
ivie_at_cc.usu.edu
Received on Wed Sep 11 2002 - 01:18:06 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:35:39 BST