| Re: To PVM or not to PVM [message #41780] |
Tue, 07 December 2004 14:44  |
Robert Barnett
Messages: 70
Registered: May 2004
|
Member |
|
|
I guess I don't know much about parallel computing. I think the better
question to ask is:
Is playing musical chairs with IDL save files a good idea, or is it
plagued with concurrency, scalability and usability issues?
Having a quick look at the FastDL site, they have made quite a nice
application which also plays musical chairs with sav files. It appears
to work fine and probably is not a bad way to approach the problem.
I like my own solution better than FastDL because I'm very vigilant
about cross platform issues. I also don't need or want a GUI because
I'll probably incorporate the distributed part into the end product.
I should probably start releasing code before I talk any more about this.
--
nrb@
Robbie Barnett
imag
Research Assistant
wsahs
Nuclear Medicine & Ultrasound
nsw
Westmead Hospital
gov
Sydney Australia
au
+61 2 9845 7223
|
|
|
|
|
|
| Re: To PVM or not to PVM [message #41861 is a reply to message #41797] |
Mon, 06 December 2004 21:06  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
Robert Barnett <retsil@zipworld.com.au> writes:
> Hi,
>
> This post is about Parallel Virtual Machine (PVM). PVM is useful for
> creating distributed computing clusters. Using clusters a high
> computational load can be split up amongst many processors on a
> network. More information about PVM is avilable from
> http://www.csm.ornl.gov/pvm/pvm_home.html.
My question is simple. Why would you spend your valuable time
changing from a system that works (I gather), to a system that you'd
have to learn from scratch, and that may or may not work? I'm not
trying to be facetious. Some people like a challenge, but there has
to be some other reward, no?
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@REMOVEcow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: To PVM or not to PVM [message #41995 is a reply to message #41780] |
Thu, 09 December 2004 04:09 |
ronn
Messages: 123
Registered: April 1999
|
Senior Member |
|
|
Hi Robbie,
I am the one that wrote the IDL to PVM interface and I use it on a
mixed network everyday. I decided to use PVM since it had a book on
the how to use it and making a dll for IDL was fairly simple.
You are right about having to compile PVM on each platform and having
to use rsh on windows. But once you do it, it is done and you can
forget about it.
As far as IDL not being multithreaded, that is true. But you aren't
doing multi thread with IDL_PVM, you are doing multitasking which works
fine. In fact using my IDL interface to PVM is actually a bit easier
than standard PVM since IDL always knows the type of data that is being
sent back and forth. In standard PVM code you have a separate call for
each data type. Finally, with IDL_PVM your slave processes can either
be IDL save files or C executables.
Ronn Kling
KRS, inc.
email: r...@rlkling.com
Use these books to upgrade your IDL skills.
"Application Development with IDL"
"Calling C from IDL, Using DLM's to extend your IDL code". SECOND
EDITION includes C++ and debugging!
"Power Graphics with IDL, A Beginner's Guide to Object Graphics",
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
