Path: g2news1.google.com!postnews.google.com!33g2000vbe.googlegroups.com!not-for-mail
From: TimDaly <d...@axiom-developer.org>
Newsgroups: sci.math.symbolic,sci.math
Subject: Re: maple is slow, pari gp even slower (on 1 topic)
Date: Sun, 17 Jan 2010 15:48:44 -0800 (PST)
Organization: http://groups.google.com
Lines: 48
Message-ID: <80f64754-8f5b-46ef-9675-e05e7a8cb039@33g2000vbe.googlegroups.com>
References: <31423165-2a11-456b-9523-706d48a03c29@e16g2000yqc.googlegroups.com>
NNTP-Posting-Host: 72.23.235.203
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
X-Trace: posting.google.com 1263772124 14899 127.0.0.1 (17 Jan 2010 23:48:44 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Sun, 17 Jan 2010 23:48:44 +0000 (UTC)
Complaints-To: groups-abuse@google.com
Injection-Info: 33g2000vbe.googlegroups.com; posting-host=72.23.235.203; 
	posting-account=m2ufuQoAAAAnP4mhpj2-bRzndif0-AIg
User-Agent: G2/1.0
X-HTTP-UserAgent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.5) 
	Gecko/2008120122 Firefox/3.0.5,gzip(gfe),gzip(gfe)

On Jan 17, 5:12=A0pm, plouffe <simon.plou...@gmail.com> wrote:
> Hello,
>
> =A0I made this simple test, computing exp(1) or E to 1 million digits
> with
> Maple 13, Mathematica 7.0 and Pari Gp 2.3.4 on a 2x4 Intel E5405
> running
> at 2.0 Ghz with 4 gigs of RAM. All installations are 64 bits on a
> Ubuntu 9.10.
>
> Maple 13 : ------------------------- 165 seconds
> Mathematica 7: ------------------ 1.11 seconds
> Pari GP 2.3.4 : -------------------- =A0936 seconds
>
> What is this!??
>
> This is bad. This means that Mathematica is =A0~150 times faster than
> Maple 13 and ~900 times faster than Pari-GP ??
>
> As far as I know, exp(1) can only be evaluated with the standard
> series,
> this means that there is a big problem with the handling of large
> numbers
> with some programs. =A0I did not try with Pifast, probably takes a very
> few
> seconds with that one.
>
> I wish that someday, some people will take the time to renew
> some libraries and/or get a connection to internet to get the
> latest news since the invention of the ENIAC...
>
> Simon Plouffe

I ran a test with Axiom and it takes about 45 seconds on
a slower, single processor box.

Are you the author of the code for Maple, MMA, or Pari?
If this is a simple test and you are aware of ways to
improve the performance, did you post improved code?

It is trivial to complain that there COULD be better code.
It is somewhat more of a challenge to write that code.
So write the improved code and benchmark your version.
I would be curious to see how much better you can do.

There is no such thing as a simple job.

Tim