On Fri, 12 Sep 2003, Sundaram A wrote:
> As this error is a mathematical error, it will affect any of the
> processes while running my own codes on OSCAR. As a Mathematiclal term
> do we need to do anythig to solve or don't need to worry about the
> error.
I think you misunderstand.
This has nothing to do with the MPI implementation, nor with OSCAR.
The test program implements a well-known integration algorithm for
calculating an approximation of pi. As an approximation, it is *not* the
exact value of pi. How close it comes to the real value of pi is simply a
function of the formula used, how many nodes you use, and how many
iterations you let it run. Look at the code, and you'll see what I mean.
At the end of the run, it displays
printf("%d, iterations: pi is approx. %.16f, error = %.16f\n",
iter, pi, fabs(pi - PI25DT);
where PI25DT is a hard-coded to the value 3.141592653589793238462643
(i.e., pi to 25 digits).
So even though this program shows an "error" calculation, it has nothing
to do with the MPI implementation, OSCAR, the numerical precision of your
hardware, nor future software that you may run -- it's simply a function
of the specific example algorithm that we chose to use in the LAM/MPI
test.
All that this test shows is that an MPI application is able to run in
parallel.
--
{+} Jeff Squyres
{+} [EMAIL PROTECTED]
{+} http://www.lam-mpi.org/
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users
