From: Sabih Gerez
Subject: summary: comparison GCL and CMUCL
Date: 
Message-ID: <sabih.773602517@tweety>
On my posting for a comparison between GCL and CMUCL (both in the public
domain, see the FAQ for this newsgroup), I received two replies with
information on the subject and one request to post a summary. So here it
comes:

* size of runtime executable: GCL about 3MB, CMUCL about 12MB.
* speed: CMUCL is faster up to a factor 5, especially for interpreted code.
* compilation: CMUCL has better diagnostics.
* debugging: CMUCL better, link with source code in compiled code.
* extensions: CLOS and CLX are built in CMUCL, to be done by you in GCL.
* installation: CMUCL is a binary distribution, GCL needs compilation.
* portability: CMUCL is available for a few platforms only (see FAQ), GCL
  for many more.

In the mean time, I have installed GCL on a SUN 4/50 and code that once was
written for Lucid on Apollo (2000 lines) compiled and executed without the
necessity for a single modification (except for the filenames that now have
a .lsp extension instead of .lisp)! The code is some factor 5 slower than the
performance of Lucid on an Apollo DN 10000 (but how do the hardwares
compare?).

Sabih H. Gerez, University of Twente, Dept. of Electr. Engineering (EL-BSC),
P.O. Box 217, 7500 AE Enschede, The Netherlands. Tel.: X-31-53-893156.
Fax.: X-31-53-334701. E-mail: ·····@nt.el.utwente.nl 
From: Ken Anderson
Subject: Re: summary: comparison GCL and CMUCL
Date: 
Message-ID: <KANDERSO.94Jul10102729@wheaton.bbn.com>
In article <···············@tweety> ·····@nt.el.utwente.nl (Sabih Gerez) writes:

  ...

   In the mean time, I have installed GCL on a SUN 4/50 and code that once was
   written for Lucid on Apollo (2000 lines) compiled and executed without the
   necessity for a single modification (except for the filenames that now have
   a .lsp extension instead of .lisp)! The code is some factor 5 slower than the
   performance of Lucid on an Apollo DN 10000 (but how do the hardwares
   compare?).

Performance isn't necessarily portable.  You should profile the program to
see where to time is going.  It may be that small differences between the
Lisp's (ie what is inlined, and what isn't) could make a significant
difference that can easily be corrected.

k
--
Ken Anderson 
Internet: ·········@bbn.com
BBN ST               Work Phone: 617-873-3160
10 Moulton St.       Home Phone: 617-643-0157
Mail Stop 6/4c              FAX: 617-873-3776
Cambridge MA 02138
USA