Robert G. Brown
rgb at phy.duke.edu
Sat May 27 12:36:02 PDT 2000
On Sat, 27 May 2000, lenins wrote:
> Hi wulfs,
> As a final year project,I am planning to write a Quantum computer
> The aim of such a simulator will be to give very accurate simulation of a
> The project calls for solving physics equations. The complexity lies in
> the fact that to calculate a function of 8 bit input, in the worst case the
> sim wolud have to execute the function 2^8 times (for each of the possible
> input cases).
> In short the characteristics are
> 1. Large number of calculations (each differing only in the input
> 2. explicit parallelism
> 3. Very limited I/O operations
> Inputs available are 25-30 Celeron 400 systems.(Networked Win-NT).Some
> experience in pvm.And some very interested guys.
> I would like to know if using a beowulf is really necessary or can i do it
> with pvm.
> If a beowulf is necessary what would we have to go through to get the wulf
> up and running.
If I understand your project's physics correctly from the description...
given that your calculation sounds like it is strongly dominated by
computation with only occasional communication between nodes (or no
communication other than starting it out, where you use many nodes to
get good statistics in an e.g. master-slave scenario) you can almost
certainly get by with any kind of cluster you like. Strategically, you
can use pvm in a master/slave layout to farm jobs out (from the master)
and collect and collate results back (on the master), or you might be
able to get by with mosix, or even just a shell script to distribute
embarrassingly parallel jobs and collect results.
25-30 Celerons should give you a fair amount of computational power
provided that your calculation isn't very poorly structured w.r.t. the
cache. Of course I'd never use WinNT for a variety of reasons, but suit
yourself. You should probably at least prototype and compare NT
execution times and linux execution times as well as stability and the
availability of useful resources (e.g. perl, various shell tools, simply
> lenin s
> Beowulf mailing list
> Beowulf at beowulf.org
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