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

RE: How to write fast Numerical Analysis code in CLISP?

   Date: Thu, 28 Apr 94 18:18:50 +0200
   From: pcolsen@super.org (Peter C Olsen)

   Matthias ---

   This is very interesting.  You are the only person who claimed that
   CLISP was faster than AKCL in executing compiled code.  Several people
   said that it might be better with interpreted code, but everyone else
   told me that I might get as much as one-to-two orders of magnitude
   increase in speed by switching from CLISP to AKCL.  Right now this is
   purely a professional problem, but it may become much more personal
   soon --- I hope to start work on a dissertation in the fall and I had
   planned to write the software in Lisp.

I just thought I'd throw in my two bits here, as someone who did my
master's thesis work in LISP.  For a LISP application, you should
write your code under whatever is easiest to use and debug.  Don't
worry about a factor of 10 in speed -- who cares when you're just
developing code?  So, in this case CLISP is a great platform -- it
runs on your laptop, your workstation, whatever you want.  It loads up
quickly, compiles quickly, etc.

When the time comes to use the code and generate the results, well,
you wrote it in Common LISP, right?  So then you re-compile it in
whatever the fastest thing around is, and run it on whatever the
fastest thing around is.  But if you're writing a significantly large
program, running the code is a trivially small part of what you're
going to do.  Developing it is most of what you'll do.