Parallel programming

[ted leslie]

one of the reasons i like C#  is that it is trying to work in this direction.
(and ironically enough, its MS that puts up a majority chunk of the $ and direction for this).
There are libraries for many languages to do light weight threading,
C in particular, fortan for more static approaches to certain parallel math hardware.
i dont think C#/.net (mono) is were you want to be now, but I do believe it will
be the first there (core to popular language wise) shortly.
For now, you might be best to use some libraries/tools for C.
(i did a bunch of this back in McMaster and Waterloo in late 80's/90's,
when the end of MHz was coming ... ha ha, we are only just now
hitting the wall now so to speak),
but to me , i am interested in tackling this problem core to the language,
and with some of it done (done right), with out me even having to know its happening,
and at other times, me designing to it.
Mono is actively doing work in this area now, some off of MS, and some its own initiative.
of course you need very light weight threads to make best use of
the common quad or octal processor units these days, and even then,
the weight of the thread can of course make the time worst then serial,
if the work needed is very little, and it doesn't make up for the over head in threading.
raytracing of course is a great one for parallel, and there is a 
parallel zip util out for linux now too, you might want to 
see what they use for lwt env.

as mentioned, cloud computing parallel is not related to SMP,
with cloud, you have to break off sizable computational pieces,
and they should have their own copy of the data (or most of it),
so web services work, but even distributing a ray tracer can too.
if you can break your sim. off in nice large chunks, cloud might be great.
There are libraries for this too of course.

-tl




On Tue, 4 Aug 2009 15:45:55 -0400
"William O'Higgins Witteman" <william.ohiggins-H217xnMUJC0sA/PxXw9srA at public.gmane.org> wrote:

> Does anyone have experience or can recommend a good resource for
> high-performance parallel programming?
> 
> Specifically, I have been asked to help one of my colleagues parallelize
> a simulation.  Ideally I would like to be able to run it on many, many
> Linux machines via Amazon Web Services or similar service, but it would
> be great to maximize an arbitrary number of cores as well as
> physical/virtual machines.  Thanks.
> -- 
> 
> yours,
> 
> William
> 
> 


-- 
ted leslie <tleslie-RBVUpeUoHUc at public.gmane.org>
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists