Any Recent Prolog Hardware?

Appeared in Volume 8/3, August 1995

vanroy@dfki.uni-sb.de
Peter Van Roy
9th March 1995

Prolog hardware design has died down since 1990. For information, see the Wonder Years article, JLP May/June 1994, which is available as:

http://ps-www.dfki.uni-sb.de/~vanroy/official_report.ps

It covers research up to the beginning of 1994, and you can see that there is very little recent work.

Building Prolog machines isn't as interesting as it once was because compiler technology and stock hardware have both improved so rapidly. Prolog is now well-understood. It's more interesting to look at more recent logic languages and to see how to support them. One key factor is that the architecture must tolerate memory latency. However, compiler technology is a more fruitful research area than architecture for these languages, since the languages are still in a state of flux.

Regarding recent work, check out Kazuo Seo's thesis on Pegasus and Pegasus-II, dated March 1993 (see reference in the Wonder Years article) and Bruce Holmer's thesis on Automatic Instruction Set Design (see http://ps-www.dfki.uni-sb.de/~vanroy/Bruce.thesis.ps). At USC, people are designing a successor to the VLSI-BAM (see http://acal-www.usc.edu).