Hi, I am attending a Parallel Processing module as a part of an Advanced Post Graduate Diploma in Software Technology. We are required to take an assignment and a project using the PVM. I would be grateful if someone could suggest a reasonable sized problem which can be solved in a month(Assignment) and a slightly complicated problem which can be taken up as a project(1.5 months) Thanks in advance, Prasanna. [ April 27, 2002: Message edited by: Prasanna Wamanacharya ]
Frank Carver
Sheriff
Joined: Jan 07, 1999
Posts: 6913
posted
0
Can you tell us a little more about this PVM? I've not heard of it personaly, and judging from the lack of responses neither has anyone else here
In a single sentence, PVM is the poor man's Super Computer. We can form a cluster of homogeneous/heterogeneous computers, and use this cluster as if were a single computer, to solve a problem involving large amount of computation. The PVM homepage is at: http://www.csm.ornl.gov/pvm The User Guide to PVM: http://www.netlib.org/pvm3/book/pvm-book.html
Frank Carver
Sheriff
Joined: Jan 07, 1999
Posts: 6913
posted
0
Looks cool. For your projects how many machines would you likely have available to play with?
Prasanna Wamanacharya
Ranch Hand
Joined: Apr 24, 2001
Posts: 143
posted
0
Well, we have about 20-25 Intel Celeron machines in one of the labs, and I guess there are about 11 project groups, so each group could get at least 2 machines. Actually the distribution is not known to us yet.
Frank Carver
Sheriff
Joined: Jan 07, 1999
Posts: 6913
posted
0
Hmm. Parallel processing with just two Celerons is not really going to gain you anything other than experience of the issues an configuration of such a system. As for a project, I've always thought that the best projects for parallel systems are the ones with lots of small loosely-coupled units of data and a large amount of essentially similar repetetive processing for them. I'd dig out one of my many still-playing-with-it simulation projects if it were me. The things below are ideas I have started at some time.
layout a map of landscape features and place a lot of simplistic "people" with "randomly" chosen goals on it to see what paths they make given that the more a path is used, the more likely peopel are to use it.
build some simple erosion or river-formation models and run them to see how the landcape forms.
Prasanna Wamanacharya
Ranch Hand
Joined: Apr 24, 2001
Posts: 143
posted
0
Thank you Frank for suggesting a couple of projects, but I would rather like to do a project which is some what related to the industry; the projects suggested by you really look like research projects. In other words, I am looking for a project which will look good on my Resume , more than that, the interviewer should be able to understand the project quickly when I explain it to him. He should not look . I guess Stock exchanges and banking projects could be good candidates.
I agree. Here's the link: http://ej-technologies/jprofiler - if it wasn't for jprofiler, we would need to
run our stuff on 16 servers instead of 3.
0
, more than that, the interviewer should be able to understand the project quickly when I explain it to him. He should not look 
.