lisp machines vs. lisp on other boxes

[ALarson@hi-multics.arpa]

Yes, my report was was indeed sloppy (It looked good to me last night ...).
The lisps involved were Franz ExCL version 2.0, and Lucid version 1.2, both
common lisp.  The franz image had flavors, their foreign function calling
mechanism, and the multiprocessing packages loaded.  Lucid was straight
common lisp (at least there were no other packages defined).  Both were
running on a 4MB main memory Sun 3/50 running diskless, with a 16MB swap
file.  I believe the Franz compile was running as a subprocess to GNU
emacs, lucid was straight from the shell.  I'm not sure what the network
traffic was at the time, although both tests were done late in the day when
traffic on our net is usually very low.  Both were run after expanding the
dynamic memory until it wouldn't get any bigger (It appears that for both
lisps the larger the dynamic memory, the faster they run).  My observations
are only for the system involved, I have not had an oportunity to test both
the lisps together on a larger memory system, and it is possible that
having additional main memory could even things out.

My results support the factor of two compilation speed that Rob
mentioned, and since I never finished compiling the files with Lucid, I
cannot comment on the execution speeds.