more benchmarks

[Billy Joel <riedesel@daneel.den.mmc.com>]

I'm confused.
I've been experimenting with AKCL and the gabriel benchmarks from
rascal.ics.utexas.edu.  I'm finding Allegro Common LISP 3.1.13 running
on a Solbourne 5/501 (sparc) to be from 2 to 4 times slower than AKCL!

What I am looking at are the benchmarks without the floating point tests.
In general Allegro does better floating point.  However I use very little
floating point in my normal work.  Allegro really seems to suffer with
the traverse and triang tests.  When I take out the floating point and
the traverse and triang tests, Allegro is 1.848 times the speed of
AKCL.  With ALL the tests Allegro is 1.47 times the speed of AKCL.
The tests without the floating point but with everything else put
Allegro at 3.833 times the speed of AKCL.  (By the floating point tests
I am including fft-mod fft and frpoly.)

I guess I just don't understand this.  Since Allegro is a native compiler
and AKCL compiles to C code I would have thought Allegro should give 
consistently better performance.  I realize these benchmarks are
a VERY limited measure of performance for these systems but surely there
should still be some kind of explanation.

Any ideas?  I'm interested.

Joel Riedesel