Path: utzoo!yunexus!maccs!sparkles!darel
From: darel@sparkles.dcss.mcmaster.ca (Darel Mesher)
Newsgroups: comp.sys.transputer
Subject: Re: Transputers in the U.S.
Keywords: Transputers, research
Message-ID: <1852@sparkles.dcss.mcmaster.ca>
Date: 26 Jan 89 14:35:58 GMT
Article-I.D.: sparkles.1852
References: <870@expya.cs.exeter.ac.uk>
Reply-To: darel@sparkles.UUCP (Darel Mesher)
Organization: McMaster University, Hamilton, Ontario
Lines: 61

In article <870@expya.cs.exeter.ac.uk> sru@cs.exeter.ac.uk (Steve Rush) writes:
>Dear Net, 
>  
>  Over here in England, we are all hooked on the Transputer and it
>seems to have been adopted by several other European establishments.
>At present however there appears to be very little happening over
>there in the U.S of A. Are you all being terribly quiet or has the
>Transputer simply not taken off yet? 
>


	I don't know specifically of the research thrusts in the States
but I can comment on the transputer history here at McMaster...

About three years ago the Computer Science and Systems department decided
to investigate the much acclaimed Inmos Transputer, so they carefully 
allocated the research money required to purchase a development board
from the much sought after and carefully guarded research accounts.  After 
several months of eager anticipation the only local Inmos sales office (and
I think the only one in Canada) finally delivered our B004 board, with firm
promises to promptly follow up with the 'C' compiler we ordered.  More weeks
passed and after experimentation with TDS and OCCAM (the folding editor is 
indeed slick) for which there are no low level routines for basic I/O and 
other standard library functions, the compiler arrived.

Examination of the compiler revealed that in fact it was a 'Pre-Production'
BETA release with cover notes to the effect that "Although this product is
a full implementation of the K&R C definition, it is extremely inefficient".
Well, the people who had heard the 10-MIP performance claims made by Inmos
for the T414 began to arrive, with 'C' benchmarks in hand to evaluate the 
system.  Needless to say, the B004 with the BETA compiler clocked performances
lower than a PC/AT class machine with an 80287.  

Interest in the transputer diminished to a point where only a handfull in
the department considered the system anything other than another overated
architecture.  Attempts at updating from the 'Pre-Production' compiler were
unfruitful with the Canadian distributors showing no interest.  The system
sat dormant with the exception of limited use for an undergraduate 
architecture course laboratory.  

Perhaps the most disheartening aspect of our involvement came after one
of the university's professors returned from sabbatical in the UK expounding
the virtues of Transputers with stories of donated software, complete support,
free updates leading to mature production versions of the dreaded C compiler;-
which by the way was not updated by the sole Canadian distributor.  When 
the possibility of purchasing any other language compilers or updates directly
from the UK is brought up for discussion, the purse strings are drawn tightly
:- proving the adage "Once bitten, twice shy!"

Perhaps the recent buy out of Inmos will produce better marketing and support
here in North America, but until things change there is little hope of 
exploratory hardware/software purchases by anyone who has had dealings with 
the old support philosophy.



Darel.

-- 
Darel Mesher				...!uunet!mnetor!maccs!darel
McMaster University			    darel@maccs.mcmaster.ca