From deleeuw Fri Feb  3 10:31 PST 1995
Return-Path: <deleeuw>
Received: from galton.math.ucla.edu by laplace.math.ucla.edu 
	(Sendmail 5.0/1.11) id AA12935; Fri, 3 Feb 1995 10:31:24 +0800
Received: by galton.math.ucla.edu 
	(Sendmail 4.1/1.12) id AA23918; Fri, 3 Feb 95 10:31:22 PST
From: Jan Deleeuw <deleeuw>
Message-Id: <9502031831.AA23918@galton.math.ucla.edu>
Subject: speed
To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List)
Date: Fri, 3 Feb 1995 10:31:21 -0800 (PST)
Cc: deleeuw (Jan de Leeuw)
X-Mailer: ELM [version 2.4 PL24]
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=US-ASCII
Content-Length: 1240
Status: RO

In view of recent discussion about the relative merits of Xlisp and
Common Lisp, and about the differences between byte-compiled and
object-compiled code, here is a tiny speed comparison using the
write-latex-table fucntion I posted recently. I have done similar
comparisons using floating point operations, and then the
results are quite different. This function is just IO and looping.

(0) All runs on galton, a Sparc 20 with 64MB and 1 processor
    Command (time (write-latex-table (make-array '(100 100))))

(1) Xlisp-Stat
	interpreted lsp 3.27 sec
	byte-compiled   0.83 sec
(2) Clisp
	interpreted lsp real 10.54 run 10.43 gc 0.26
        byte compiled   real  6.36 run  6.24 gc 0.24
(3) cmucl
	interpreted lsp real 12.10 run 11.61 gc 1.53
        object code     real  4.74 run  3.45 gc 1.06
(4) gcl
	interpreted lsp real  5.73 run  3.27
        object code     real  3.78 run  1.55

I'll do Xlisp-Stat vs MCL when I get home (on the PowerBook 170).
I cant do PowerMac, because my Powermac is in LA, and I am
in NC.
-- 
Jan de Leeuw; UCLA Statistics Program; UCLA Statistical Consulting
US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90024-1554
phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu