This is the mail archive of the
gsl-discuss@sourceware.org
mailing list for the GSL project.
Re: OpenMP GSL
- From: "Andrew W. Steiner" <awsteiner at gmail dot com>
- To: "Jochen Küpper" <kuepper dot jochen at googlemail dot com>
- Cc: "GSL Discuss Mailing List" <gsl-discuss at sourceware dot org>
- Date: Mon, 14 Apr 2008 15:08:53 -0400
- Subject: Re: OpenMP GSL
- References: <5644665F-AAC2-469C-920D-6B13DBE87085@googlemail.com>
Exciting thought, but I'd lean towards a separate package. In any case
"parallelization of GSL" is kind of a loaded term. It's not clear to
me that many of the routines presented in GSL are easily
"parallelizable" per se?
On Mon, Apr 14, 2008 at 2:46 PM, Jochen Küpper
<kuepper.jochen@googlemail.com> wrote:
> Hi,
>
> I was wondering whether there are any plans for the parallelization of GSL?
>
> For example, I found this article on "Parallelization of GSL: Architecture,
> Interfaces, and Programming Models" form 2004:
>
> https://commerce.metapress.com/content/jtp50vjc1e3gwane/resource-secured/?target=fulltext.pdf&sid=medqeqrui2yyng45okahx345&sh=www.springerlink.com
>
> I also found a message by Brian Gough from 2002 stating that it would not
> fit the design of GSL (Message-ID: <15656.48877.30970.694792@debian>).
>
> Would there be any objections if one started to a few include OpenMP
> directives in the code? They would not at all interfere with normal usage of
> GSL in single-threaded calculations. If they should even be optional for the
> use of GSL in OpenMP calculations, they could be flagged by a run-time or
> compile-time (preferred for performance) switch.
>
> Thoughts?
>
> Greetings,
> Jochen
> --
> Einigkeit und Recht und Freiheit
> http://www.Jochen-Kuepper.de
> Liberté, Égalité, Fraternité GnuPG key: CC1B0B4D
> Sex, drugs and rock-n-roll
>
>
>