Issue 599 in fds-smv: FDS - Pointer not associated with a target error

[codesite...@google.com]

Comment #2 on issue 599 by pfh...@ameritech.net: FDS - Pointer not
associated with a target error
http://code.google.com/p/fds-smv/issues/detail?id=599

After further investigation, I was able to run the case file with out
getting the
error using an FDS built with the following compiler flags which are for
the most
part the same as the intel64 osx flags in the makefile.

intel_linux_mpich_64 : FFLAGS = -O3 -m64 -heap-arrays -axSSSE3
-static-intel
intel_linux_mpich_64 : CFLAGS = -O3 -m64 -Dpp_noappend

However, the run stalls as shown below. Viewing the system monitor, no
processor
activity for FDS is apparent.

[Paul@Xeon PGCatenary5-test]$ mpiexec -n 2
/FDS2/PGCatenary5-test/fds5_2_mpi_linux_hartosx
/FDS2/PGCatenary5-test/pressure_rise.fds
Process 0 of 1 is running on Xeon
Process 1 of 1 is running on Precision
Mesh 1 is assigned to Process 0
Mesh 2 is assigned to Process 1

Fire Dynamics Simulator

Compilation Date : Wed, 31 Dec 2008
Version : 5.2.5 Parallel
SVN Revision No. : 2968

Job TITLE : Pressure Rise in a Sealed Tunnel, SVN $Revision$
Job ID string : pressure_rise

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

[codesite...@google.com]

Comment #3 on issue 599 by pfh...@ameritech.net: FDS - Pointer not

associated with a target error
http://code.google.com/p/fds-smv/issues/detail?id=599

On another note, I was also experiencing crashes running Cases with a large
number of
cells per mesh (1,600,000 cells). Using the compiler flags above resolved
the issue.

[codesite...@google.com]

Updates:
Status: Started

Comment #4 on issue 599 by mcgratta: FDS - Pointer not associated with a
target error
http://code.google.com/p/fds-smv/issues/detail?id=599

When it rains it pours -- I ran into the same bug involving the
MASS_FILE=.TRUE.
It's nothing more than a problem where FDS is using the array YY (species
mass
fractions) which has not been allocated because there are no species, not
even
mixture fraction. That's probably why we haven't seen it up to now.

In diagnosing that problem, I then ran into your second one. It looks like
it's a
bug introduced after the official release of 5.2.5 and should be easy to
fix. Hold
off on trying to sort this out because I should be able to do it this
morning.
Thanks.

[codesite...@google.com]

Comment #5 on issue 599 by pfh...@ameritech.net: FDS - Pointer not

associated with a target error
http://code.google.com/p/fds-smv/issues/detail?id=599

I think it may be the same problem. The attached file also involves the
MASS_FILE=.TRUE. I use the pressure_rise.fds test case for multi-mesh mpi
tests. The
attached file is a simplified version of this test case.

[codesite...@google.com]

Updates:
Cc: drjfloyd

Comment #6 on issue 599 by mcgratta: FDS - Pointer not associated with a
target error
http://code.google.com/p/fds-smv/issues/detail?id=599

You're one step ahead of me. Yes -- the bounds of some of the arrays
associated with
MASS_FILE had to be changed, but we forgot to modify the MPI calls
involving these
same arrays. I'll post new code this afternoon. In the meantime, don't use
MASS_FILE.

Jason, FYI, the arrays in question are MINT and MINT_SUM. In main_mpi.f90,
the MPI
calls involving them assume the old array bounds.

[codesite...@google.com]

Updates:
Status: Fixed

Comment #7 on issue 599 by mcgratta: FDS - Pointer not associated with a
target error
http://code.google.com/p/fds-smv/issues/detail?id=599

I think I fixed both bugs. Could you retry your calc with the latest and
let me
know. Thanks.

[codesite...@google.com]

Comment #8 on issue 599 by pfh...@ameritech.net: FDS - Pointer not

associated with a target error
http://code.google.com/p/fds-smv/issues/detail?id=599

I recompiled with revision 2994. It appears as though the problem has been
resolve
since I no longer receive the error.

[codesite...@google.com]

Updates:
Status: Verified

Comment #9 on issue 599 by mcgratta: FDS - Pointer not associated with a
target error
http://code.google.com/p/fds-smv/issues/detail?id=599

Great, thanks for checking.