Dartmouth repository for parallel-I/O code examples.

Example name: CFS3D

Description:
   This is a synthetic code meant to be similar to a real application
called ARC3D, in terms of its I/O. 

I/O usage:
   It outputs a 3-d matrix that is spread across distributed memories
into a single sequential file, doing the I/O in parallel to a CFS
file. All I/O is done at the end of the execution. The arrays are
decomposed across processor memories, with some overlap. Overlapping
parts of the file are just overwritten.  One file contains three 3-d
arrays called X, Y, and Z (in "Plot3d format"); another file contains
a 4-d array called Q. Node 0 creates and preallocates the file. Each
processor writes its own subcube of the matrix, but that subcube is
not contiguous in the output, leading to an interleaved pattern.

Platform: iPSC/x with CFS file system

Language: FORTRAN (77?)

Code author(s): 
	Steve Ryan
        NASA Ames NAS facility
        ryan@nas.nasa.gov

Contributed by: David Kotz, Dartmouth College, dfk@cs.dartmouth.edu
Date contributed: 7/28/93

-------------------------------------------------------
Notes from the author:

The files in this directory are to be released to anyone who wants
to use them as examples for testing of CFS I/O, or to find improved
ways of doing this type of I/O.

The cfsioS2.f and cfsXS.f files contain the I/O parts of the code.
They allow a grid which has been distributed across several nodes to
be reassembled as a single array in a single file on CFS.  This file
could then be moved to other machines for use in visualization or 
other post-processing.

Edit the file names which are hard-coded in the "open" statements of
these routines.  para.h contains some important parameters, and may 
be changed as necessary.

go4 is a simple script to run the example on 4 processors.

Any suggestions for improving the I/O performance are welcome.  The
calculations made in the code are not meant to be relevant.  This
is just a dummy code for trying out the message passing and disk I/O.

Steve Ryan
(415) 604-4496
ryan@nas.nasa.gov

More information can be found in the following paper:
@InProceedings{ryan:navier,
  author = {J. S. Ryan and S. K. Weeratunga},
  title = {Parallel Computation of {3-D Navier-Stokes} Flowfields for
  Supersonic Vehicles},
  booktitle = {31st Aerospace Sciences Meeting and Exhibit},
  year = {1993},
  address = {Reno, NV},
  note = {AIAA Paper 93-0064.},
  URL = {http://www.nas.nasa.gov/HPCC/Pubs/ryan/AIAA93-0064.html},
  keyword = {parallel application, CFD, parallel I/O, pario bib},
  comment = {This paper goes with the ryan:cfs code example.}
}