[Beowulf] massive parallel processing application required
Peter St. John
peter.st.john at gmail.com
Wed Jan 31 10:36:57 PST 2007
I advocate building your own data, which is trivial in mathematics
applications. One of the first uses for distributing hard computations to
volunteers on the net with idle CPU time was number theory (primality
testing, finding primes, for cryptography, for example).
A site that provides various source code to do some of these things is
ECMNET ("Elliptic Curves Method'),
It isn't necessary to do any number theoretic research yourself, but you'd
want to be comfortable with sentences like "use <this algorithm> to factor a
candidate Fermat Number". In this example, a Fermat Number is a particular
number that is "probably" prime. It's hard to factor, to find out if it is.
If it is, you can use it to generate public keys for cryptography. If it
isn't, somebody will read your mail. So you have to look up the formula for
Fermat numbers and appy the algortithm to try and factor it. You don't need
to do any math yourself besides elementary algebra, and there are software
packages for everything.
An example from the site: "Peter Montgomery found in November 1995 a factor
of 47 digits of 5^256+1". The exponential thing, a huge power of a prime,
plus one, would be the possible prime; it was proven not to be prime, by
finding a 47 digit prime factor. Nowadays 47 digits is chump change; I don't
know the current records. But you can burn up all your nodes by asking each
of them to do a few factorizations of this nature, and you can become famous
by finding a new largest prime.
A somewhat prettier site about Elliptic Curves is
There are probably contributors to the Cunningham Project (mentioned by
ECMnet) that would love to help you implement the app on your cluster, in
exchange for access to your cluster for their apps, which in this case would
amount to the same thing.
On 1/31/07, Mitchell Wisidagamage <06002352 at brookes.ac.uk> wrote:
> Thank you very much for the fire dynamics idea. I will have a look at it.
> I did try to contact many e-science projects including some researchers
> at Oxford. But I got no reply. Then I went to get some contacts from a
> tutor who worked at a e-science project himself. He told me people,
> especially scientists are "very jealous" of their data. And not replying
> is a kind way of saying "no". And there's the problem of "who's this guy
> wanting my data", "what will he do with it?".
> I have given up the e-science idea. Now looking for other real world
> John Hearns wrote:
> > Mitchell Wisidagamage wrote:
> >> Hi all,
> >> As part of my dissertation, I'm looking for "raw data" which will be
> >> used for massive parallel processing using Beuwulf cluster (with the
> >> use of PVM or MPI). I tried looking for e-science raw data (and the
> >> computations required on it) such as bioinformatics, fluid dynamics,
> >> etc. but without any luck.
> >> Anyone has any idea of getting some raw data so I can give compute
> >> intensive "work" to the nodes?
> > Mitchell,
> > how about running the NIST Fire Dynamics simulation?
> > http://www.fire.nist.gov/fds/
> > It simulates the spread of smoke and fire in buildings.
> > There are some sample input models for download.
> > The Smokeview program visualizes the output, which will be a nice
> > demonstration for your tutor.
> > But why not just go across to the Oxford E-science centre?
> > I know for sure they have one cluster there for handling large datasets!
> > Ask them for help in getting a suitable dataset for your project.
> > Drop me an email if I can give you any advice, you're in my neck of the
> > woods.
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Beowulf