Parallel Beta Test

[Kevin McGrattan <kevin.mcgrattan@nist.gov>]

All

I've received a few inquiries about a parallel version of
FDS. We're developing
this capability for use in our World Trade Center
investigation. For those of
you who would like to try it, I have set up a directory on
the FTP site

ftp://ftp.nist.gov/pub/bfrl/mcgratta/FDS4

containing the source code, user's manual, and various other
things.

For those of you who don't know anything about parallel
processing, do not try to
do this just yet. We are going to have to go through a
period of testing and
streamlining before the code is readily usable on PCs. We
have done most of the 
development on a Linux cluster running LAM/MPI, but I have
run a short program on
several PCs using MPICH.

For those wanting to try this, you will have to set up your
network to run MPI.
Most will have to compile the code using the appropriate
libraries. I have a 
version of FDS compiled with LAM for a Linux cluster, and
MPICH for a Windows
cluster. For everything else, you're on your own. Note that
there is a short
program called hello.f in the MISC directory. I encourage
anyone trying to run the
code in parallel to compile and run this ``Hello World''
program first. If it 
doesn't run, nothing else will.

We're calling this version FDS4, although the beta test will
be tagged with 3.99
until we officially release. The source code has two
different main.f routines. 
main.f is a single processor version of FDS, allowing one to
compile and run the
code the usual way. For those running in parallel, use
main_mpi.f instead. Read
the User's Guide for more details on compiling and running.
Also note that we had
to make a few changes to the PARTicle namelist group, so
watch out for that.

Let me know how it goes for you brave souls who try this
first. Hopefully we can
work out the kinks and move the development along quickly.


In other news...

A small software company called Thunderhead Engineering
(thunderheadeng.com) has
won an SBIR (Small Business Innovative Research) grant from
the NSF (National
Science Foundation) to build a GUI (Graphical User
Interface) for FDS. Although
NIST is not funding this effort, we have agreed to cooperate
with the company
by providing them with whatever information they need to
write their software.
Ultimately, the company will sell the software to interested
users. Our plans
are to continue developing FDS and Smokeview and releasing
them publicly. The
GUI will be an extra program that one can purchase that will
simplify the writing
of input files.

I know of various engineering firms that have developed
different ways of streamlining
the writing of input files, especially in regards to
building complicated geometries.
As the Thunderhead effort is in mid-stream, anyone is
encouraged to throw ideas
their way.

I also was informed by Bryan Klein of the Western Fire
Center (bryan@westernfire.com)
that he has developed some routines that will convert output
from a solids modeling
package called VectorWorks into input for FDS. Anyone
interested may contact
Bryan directly.


That's all for now. Hope all is going well with the present
release of FDS/SV. Let
me know how the parallel version works.

Kevin


-- 
Kevin McGrattan
National Institute of Standards and Technology
100 Bureau Drive, Stop 8663
Gaithersburg MD 20899-8663

Telephone: (301) 975 2712
FAX:       (301) 975 4052