Windows package built w/ CCL



On Fri, 4 Mar 2011, Barton Willis wrote:
> maxima-bounces at math.utexas.edu wrote on 03/04/2011 10:47:54 AM:
>> I HAVE tested GCL and CCL on Windows, running Maxima 5.23.2, and
>> there is a very significant difference between the two in
>> terms of running speed.
> For the testsuite (using a Core 2 6400 (2.13 GHz) and 2 GB RAM), CCL is
> about 30% slower
> than GCL. Likely the testsuite is a poor speed benchmark.
My results differ (ASUS eeePC netbook, 1.66GHz Atom N450, 2Gb RAM, 32-bit 
Gentoo Linux):

sbcl-1.0.45    165
cmucl-20b      172
gcl-2.6.8_pre  172
clozurecl-1.6  322
ecl-11.1.1     344
clisp-2.49     552

(this is real time in seconds). So, ccl is nearly 2 times slower than gcl.

Of course, the testsuite is not a universal benchmark. The speed for some 
problem depends on how much work is about multiple-precision arithmetics, 
and how much work is pure symbolic. Different lisps have different 
implementations of long-integer and high-precision float arithmetics, and 
the speed of this implementation is an independent parameter, nearly 
unrelated to the speed of ordinary list processing (used for symbolic 
calculations).

Andrey