SPARC times

[boyer@cs.utexas.edu]

Below are runs on a Sun-4/280, showing Sun/Lucid and AKCL times for
the same test and same compiler settings as in the last couple of
messages.  Is it fair to time Lucid 2.1.3, or is 2.1.3 too old?  It's
the only version I have on this Sun-4.

These times below seem to show AKCL with a 1 microsecond funcall
(after subtracting out the do-cost) and Sun/Lucid with a 1.5
microsecond funcall.

As you say, this microtiming stuff is extremely treacherous.  The main
point of my publicizing these tests was just to point out that AKCL
now really has a much better and really competitive funcall IF you
make the right proclamations.  I hope no one will misconstrue these
runs as a sweeping, allegedly scientific claim about the speeds of the
various Lisp compared.  It's just a little evidence.

AKCL (Austin Kyoto Common Lisp)  Version(1.241) Fri Oct 20 06:53:52 CDT 1989
do-cost
real time : 0.367 secs
run time  : 0.300 secs

regular-call-cost
real time : 0.883 secs
run time  : 0.867 secs

funcall-cost
real time : 2.367 secs
run time  : 1.317 secs
NIL

;;; Sun Common Lisp for the Sun-4, Development Environment 2.1.3
;;; 11-Jan-89

do-cost
Elapsed Real Time = 0.60 seconds
User Run Time     = 0.45 seconds
System Run Time   = 0.00 seconds
Total Run Time    = 0.45 seconds

regular-call-cost
Elapsed Real Time = 2.05 seconds
User Run Time     = 1.52 seconds
System Run Time   = 0.02 seconds
Total Run Time    = 1.54 seconds

funcall-cost
Elapsed Real Time = 5.51 seconds
User Run Time     = 2.05 seconds
System Run Time   = 0.17 seconds
Total Run Time    = 2.22 seconds