Newsgroups: comp.lang.apl
Path: watmath!watserv2.uwaterloo.ca!torn!cs.utexas.edu!zaphod.mps.ohio-state.edu!darwin.sura.net!haven.umd.edu!uunet!s5!hui
From: hui@fid.morgan.com (Roger Hui)
Subject: Re: performance update, apologies offered
Message-ID: <1993Jan9.011248.21443@fid.morgan.com>
Keywords: performance, j
Organization: Morgan Stanley & Co., New York, NY
References: <1993Jan7.211247.3714@nic.csu.net> <1993Jan8.151656.15252@fnbc.com>
Date: Sat, 9 Jan 1993 01:12:48 GMT
Lines: 27

Joseph Alotta writes:

> j is still some 37% slower, but i think the difference can be 
> partially accounted for by the fact that j allows complex numbers
> and z does not, and j allows n-dimensional arrays where z allows
> up to 3 dimensions.
> 
> amy{jja}15: cat ztest
> a=iota(1e6);
> +/a;
> amy{jja}16: cat jtest
> a=. i. 1e6
> +/a
> amy{jja}17: time zlang ztest
> 5.000005e+11
> 3.5u 0.8s 0:04 95% 0+0k 2+0io 0pf+0w
> amy{jja}18: time j < jtest
> 4.999995e11
> 2.2u 1.1s 0:03 96% 0+0k 0+0io 0pf+0w
> J6   Copyright (c) 1990-1992, Iverson Software Inc.  All Rights Reserved.
> 
>    1.1 % 0.8
> 1.375

It seems to me that this benchmark compares the time required to do 
the sum and to load/initialize the interpreters, and therefore may not 
necessarily reflect the relative performance of the two interpreters.
