[Beowulf] Pi calculator/RAID accross all nodes/Mosix vs. OpenMosix
Robert G. Brown
rgb at phy.duke.edu
Sun Feb 27 23:28:51 PST 2005
On Sun, 27 Feb 2005, Jake Thebault-Spieker wrote:
> A couple of questions.
> 1. Is it possible to use the HD in the nodes on the cluster as one
> HD(kind of like an extended RAID array)?
Yes. Google for PVFS.
> 2. Does anybody know of a program that will calculate pi, one digit at a
> time, infinitely that will run in parallel?
I don't know about one that will compute an infinite number of digits in
PI, but the computation of PI via the arctan series is trivially
partitionable in a variety of ways. You'll spend more time working to
sum and align the digits you get (as they obviously will have to be
obtained and manipulated piecewise as strings) than you will doing the
computation per se. It actually sounds like a decent exercise, as the
carry from small digits may have to propagate iteratively back to larger
ones as you extend the computation farther and farther.
However, it ALSO sounds like one of those problems where parallization
may not do too well trying to beat a well-written serial version.
Also, IIRC there are example programs for computing pi in parallel in
lam and mpich, but I don't think they are geared for returning all the
digits as a digit string.
You might look at the following:
for some of many online articles on pi and its computation. Google is
> 3. What is the difference between Mosix(www.mosix.org) and
I don't know, don't use Mosix. But somebody on list probably does.
> I'm in the process of reading "Engineering a Beowulf Style Computer
> Cluster" by Robert Brown. I like it a lot and it contains lots of
> information. Thanks Mr. Brown! ;-)
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
More information about the Beowulf