lisp machines vs. lisp on other boxes

["David M. Siegel" <dms@hermes.ai.mit.edu>]

   Date: Tue, 26 May 87 13:46 EDT
   From: Daniel L. Weinreb <DLW@alderaan.scrc.symbolics.com>
   Resent-From: CMP.SLUG@r20.utexas.edu
   Resent-To: SLUG:;

   Having certain hardware that's particuarly useful for Lisp does not, by
   itself, spoil any of the performance gains that are claimed by
   architectures that claim to be "RISC".  For example, the 3600
   tag-checking and EGC hardware are very simple, very small, and don't
   slow down the critical paths of the processor at all.  Note that the
   Berkeley group themselves are working on something they describe as a
   RISC machine specialized for Lisp.

A problem that Symbolics, and any company that relies on their own
custom hardware, has is that they are in constant competition with
hardware houses. For example, Sun almost doubled the speed of their
Sun-3 by going from a 16 M Hertz 68020 to a 25 M Hertz part.  Sure,
they had to do some redesign of their CPU board, but it was Motorola
that did all the hard work for them.

For Symbolics to get speed improvements, they need a massive in-house
design effort. This in part explains why over a many year period they
have hardly speeded up their original 3600, while the speed
improvements from a Sun-1 to a Sun-3 have been huge.

If I recall correctly, Computer Vision (a CAD company) used their own
hardware for many years. Although they were a leading vendor in the
CAD market, other companies caught up to them by using off the shelf
workstations. Computer Vision built their own computers because they
felt CAD could only be done fast with specialized architectures.
Lately, they have ported their software to some generic Unix box.