I tried using SBCL 1.0.13 for Windows XP and CVS Maxima. I get
3,397,963 calls to great. The calls to great consume about 12 seconds
out of 99. GCL runs the test suite in about 61 seconds, and great uses
about 24% of the run time.
I used a pedestrian method; let me know if it's wrong or silly:
(defvar *calls-to-great* 0)
(defvar *great-time* 0)
(defun great (x y)
(let ((start))
(incf *calls-to-great*)
(setq start ($elapsed_real_time))
(prog1
(great-xx x y)
(incf *great-time* (- ($elapsed_real_time) start)))))
(defmfun great-xx (x y)
...)
Evaluation took:
98.86f0 seconds of real time
72.21875f0 seconds of user run time
25.328123f0 seconds of system run time
[Run times include 3.092f0 seconds GC run time.]
0 calls to %EVAL
0 page faults and
4,721,320,584 bytes consed.
(%o0) done
(%i1) :lisp(print *calls-to-great*)
3397963
3397963
(%i1) :lisp(print *great-time*)
11.9869999999997
11.9869999999997
(%i1) build_info();
Maxima version: 5.14.0cvs
Maxima build date: 18:57 4/3/2008
host type: i686-pc-mingw32
lisp-implementation-type: SBCL
lisp-implementation-version: 1.0.13
Barton
-----maxima-bounces at math.utexas.edu wrote: -----
>> We have a big discrepancy on the number of calls to 'great.' You got
>> 1,661,359, and I got 3,035,005. What's the story? Let me double check
>> to see if I was doing something wrong or nonstandard.
>
>Don't know, but I'm not sure I trust the results from cmucl either. The
>profiler complained about some things, and then when I asked for the
>results, it complained about break-call not having any profile info
>structures associated with it.
>
>I should try again too, more carefully.
>
>Ray
>_______________________________________________
>Maxima mailing list
>Maxima at math.utexas.edu
>http://www.math.utexas.edu/mailman/listinfo/maxima