In this inquirey I wish to exclude so called 'lisp machines'
and restrict the opinions to machines found in the typical
laboratory environment (Sparc, SGI,HP,DEC,Intel, ..etc.)
Let's also exclude MPP platorms such as the CM-5 running
*Lisp and just stick to high-performace desktop scientific
workstation systems.
Question: Which ones run common Lisp the best? Which common Lisp
is the best for a given architechture?
I am particularly interested in the free implimentations CMUCL and GCL.
Go ahead, start a raging flame war or religous crusade. We haven't had
one this group for a while. I don't count that "why people prefer C"
argument, it's apples and oranges IMHO.
I really want to know, I've got some serious work to do.
--
-----------------------------------------------------------------------------
Ray Hann |
NASA Lewis Research Center |
Cleveland, Ohio 44135 | email: ······@hyperthink.lerc.nasa.gov
-----------------------------------------------------------------------------
Deep Magic in Scheme
==>(set! x (eval '(define (set! a b) (sows_ear->silk_purse C++-program))))
-----------------------------------------------------------------------------
In article <··········@sulawesi.lerc.nasa.gov> ······@hyperthink.lerc.nasa.gov (Ray Hann) writes:
Question: Which ones run common Lisp the best? Which common Lisp
is the best for a given architechture?
I am particularly interested in the free implimentations CMUCL and GCL.
As you probably know, the price/performance leader among workstations
changes almost daily now. It also depends on whether you are
interested in integer or floating-point performance or graphics, and
what kinds of discounts you can scrounge. So you have to shop around.
CMU CL doesn't run on every machine, so it limits your choices. It
has been run longest and most intensively on MIPS-based Decstations
under Mach, but that isn't a good option now. Of popular machine/OS
combinations, the most-used choice is Sparc/SunOS. Other things being
equal, CMU CL will be better debugged and better optimized on that
platform. The other viable option as of today is the HP 700 series
under HPUX.
I know much less about GCL, but it runs on more machines and its speed
is mostly determined by the underlying machine's speed and the quality
of its C compiler.
-- Scott
===========================================================================
Scott E. Fahlman Internet: ····@cs.cmu.edu
Principal Research Scientist Phone: 412 268-2575
School of Computer Science Fax: 412 268-5576 (new!)
Carnegie Mellon University Latitude: 40:26:46 N
5000 Forbes Avenue Longitude: 79:56:55 W
Pittsburgh, PA 15213 Mood: :-)
===========================================================================
From: Brian Leverich
Subject: Re: Which machines are best for common lisp?
Date:
Message-ID: <389q45$3qr@rand.org>
Ray Hann (······@hyperthink.lerc.nasa.gov) wrote:
: Question: Which ones run common Lisp the best? Which common Lisp
: is the best for a given architechture?
"Best" depends on what you want to do.
In my case, I want to build portable and freely distributable
applications for use at RAND and at our DoD clients. When _that's_
your primary goal, it's hard to beat Linux/GCL. (Actually, we use the
suite of Linux/GCL/CLX & CLOS/Garnet/ROSS simulation language.) GCL
runs on everything and Linux/GCL can go on any modern PC that's been
beefed up with enough DRAM.
Linux/GCL also probably wins if "best" means most "lisp cycles per
dollar", although a SPARC/CMUCL combination might be a close second
if portability isn't a secondary goal.
If you want flat maximum number of "lisp cycles per second", then you
probably want one of the high-end workstations with a good commercial
Lisp. Other goals might lead you to other solutions. Cheers, B.
---------
Dr. Brian Leverich
Information Systems Scientist
RAND, Santa Monica, CA 90407-2138
········@rand.org
Ray Han asks:
In this inquirey I wish to exclude so called 'lisp machines'
and restrict the opinions to machines found in the typical
laboratory environment (Sparc, SGI,HP,DEC,Intel, ..etc.)
Let's also exclude MPP platorms such as the CM-5 running
*Lisp and just stick to high-performace desktop scientific
workstation systems.
Question: Which ones run common Lisp the best? Which common Lisp
is the best for a given architechture?
Good idea for a thread! I haven't done any extensive testing, nor have I
ever seen any published, so this is just anecdotal. However, I have to say
that I am *VERY* happy with my current LISP environment, and I'm doing some
pretty compute-intensive research work in it.
I'm running Franz Allegro CL 4.2 on an SGI Indigo^2, with the 150MHz R4400
CPU with 96M ram. I think the most important thing when developing
prototype lisp systems is to have enough memory so that no swapping goes on.
I recently upgraded from 64M and the difference was just amazing. GC's are
now basically instantaneous. My development environment is FSF emacs, with
Allegro's hooks -- instant access to CLtL2 and Franz's docs, with apropos,
incremental compilation, Meta-. works well on CLOS objects and methods, and
ACL's trace now (finally!) has a decent interface to the CLOS stuff, too.
I prefer SGIs to Suns for lots of fairly idiosyncratic reasons above and
beyond the better bang for the buck (e.g. lots of MolBio software I use runs
only on them), and I haven't really given Harlequin a fair evaluation, but I
love working in this environment.
I'm very curious to hear about other people's preferences and experiences,
and a serious competitive analysis would be really invaluable.
Larry
--
Lawrence Hunter, PhD.
National Library of Medicine
Bldg. 38A, 9th floor
Bethesda. MD 20894 USA
tel: +1 (301) 496-9300
fax: +1 (301) 496-0673
internet: ······@nlm.nih.gov
encryption: RIPEM via server; PGP via "finger ······@work.nlm.nih.gov"