🚀 go-pugleaf

Has anybody done any crude research on the MIMD distributed memory
market?  What are the total number of sites in the world if
you add up the transputer boxes, the Intels and the Ncubes?
What are the figures for each?  What is the break-down in the US?

My guess is that world-wide, there are far more transputer sites
than Intel/Ncube sites combined.

I think the reasons for this are:
a) It is easy (trivial) to build a multicomputer from transputers.

therefore,

b) Transputer machines come in any size.
c) Transputer machines are cheaper.
d) There are lots of transputer vendors.
e) The little guy can afford a small transputer machine.
f) The h/w engineer with time on his/her hands can build his/her own.

These reasons tend not to apply to the latest, hottest microprocessor
from the big 3.  Of course, there is a certain part of the market
that must have the latest, hottest microprocessor and will pay
the money to get it.  But it is my hypothesis that the transputer
will continue to be the MIMD distributed memory part for the masses
until there is a competitor with a similar package.

--Greg