The Connection Machine

by W.E. Hillis

The Connection Machine describes a fundamentally different kind of computer. It offers a preview of a parallel processing computer that Daniel Hillis and others are now developing to perform tasks that no conventional, sequential machine can solve in a reasonable time.

W. Daniel Hillis is a founder of Thinking Machines Corporation where he is engaged in building connection machines as a significant step toward real thinking machines. The Connection Machine is included in the Artificial Intelligence series, edited by Patrick Winston, Michael Brady, and Daniel Bobrow.

Genres: Computers, Computer Science, Technical, Artificial Intelligence, Programming

208 pages, Paperback

Published February 15, 1989

Reviews

[Jon Gauthier · Rating 4 out of 5]

Really enjoyed the Lisp section in particular (enough to write an article about beta reduction!). Skimmed through much of the discussion on hardware as this subject is not of interest to me.

[Daniel · Rating 4 out of 5]

Ironically, one of the most perspicuous technical books I've read, even though it is about such an intuitively "lofty" concept as a supercomputer. The Lisp chapter alone, with it's treatment of xectors, was, so to speak, "worth the price of admission".

[Rainer Joswig · Rating 3 out of 5]

The book is only a very short overview about the Connection Machine. He got his PhD for his actual accomplishments and not so much for the contents of his thesis.

[Joshua]

We would indeed like to make a thinking machine. "Someday, perhaps soon, we will…"

And perhaps someday I will read this. But I don't think it would do me any good right now. And why would I do something that won't benefit me, or at least benefit someone else?

[Peter Aronson · Rating 4 out of 5]

Short and well written, this book describes the architecture and programming model of The Connection Machine, an early example of massively parallel processing. In 2016, this book is something of a historical artifact, as computer and software design went in a different direction, with parallelism being provided by huge data centers full of cheap, commodity, multi-core processor-based systems, but it shows a possible different direction that computer design could've gone. Hillis' last chapter where he discusses parallels (and lacks of parallels) between computer science and physics seems like it should still apply today (although he missed Chaos Theory, which appears to apply to both). I will note that the programming sections assume reasonably familiarity with some form of LISP, and I am rather glad I decided to mess around with Clojure last year, as otherwise it would have been tough sledding with only my 35-years-ago 1/3 of semester of Common LISP.