CL Benchmarks Using AutoClass II: Revisited

[Moon@STONY-BROOK.SCRC.Symbolics.COM (David A. Moon)]

   cpu  %-cpu  elapsed  %-elap  configuration
                 36.15    15.5  Symbolics MacIvory (20mb RAM; FPA;
	                           Genera 7.3I)

There are two models of MacIvory and you didn't specify which one this was.
>From the slow speed I would guess it was a model 1.  If you can get your
hands on a MacIvory model 2 running Genera 8.0 you might want to run the
test again; I'd expect the newer software and newer hardware to produce
a dramatic improvement, probably a bit over a factor of 2.

Also, 20MB RAM is an unusual configuration for a MacIvory, are you sure
that number is accurate?  From the other table entries it looks like you
count 4 bytes per word for Symbolics machines, so a typical MacIvory
configuration would be 10.25MB, using one so-called 16MB memory board.
(The difference is because the tag and error correction bits are being
counted in the 16MB figure, also a small portion of that board's memory
is unavailable for Lisp memory because it's used for something else.)
You might have had two 16MB memory boards and rounded down to 20MB, or
computed the memory size for the MacIvory in a way that's not consistent
with the computation for the other Symbolics and non-Symbolics machines.
I just wanted to make sure all the numbers are computed the same way.