gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Gcl-devel] Performance


From: Camm Maguire
Subject: [Gcl-devel] Performance
Date: 10 Jun 2002 18:56:30 -0400

Greetings!  Has anyone noticed that a maxima make test takes about 15%
longer with gcl-2.5.0 that with gcl-2.4.3?  I'm guessing the larger
footprint of the added functions combined with the original number of
pages allocated by default, leading to more gc.  We might want to
setup some of the new stuff to autoload instead of compiling it in.

Hope to get to this, but if anyone else wants to investigate
... great!

BTW, a crazy idea came the other day -- what if the gcl compiler could
detect at runtime whether blas libraries were available, and compile
floating point matrix code submitted in lisp to blas calls?  Its
crazy, but orders of magnitude in performance can be had transparently
by using freely available optimized blas libraries, and I think
floating point is a traditional bottleneck of lisp.  Could be a very
neat feature if it was possible.  

And again, it would be nice, though its not high priority now, if we
could import a lisp benchmark suite into the gcl tree.  I saw an old
one by gabriel in the ecls distribution.  Anyone care to check it out? 

Take care,

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah



reply via email to

[Prev in Thread] Current Thread [Next in Thread]