Resume parallel simulation

[Michele]

Resuming stopped simulations has problems at large meshes.
I''m actually working on a simulation with 300''000 cells, equally split in 2 meshes. The simulation runs in parallel mode (I have a dualcore CPU).
After stopping the simulation, the "resume parallel simulation" doesn''t work properly.
When doing the "resume parallel simulation", the "FDS simulation" windows opens properly and everything seems right. The "Time Elapsed" clock continues to increase , but the simulation does not restart, i.e. no further time steps are shown. I waited for 50 about minutes (confirmed by the Elapsed Time clock) and then had to close pyrosim manually.
This doesn''t happen when doing the non parallel "resume simulation", which restarts after a couple of seconds (even if applied to a previously parallel running simulation)
I tested both the 32 and 64 bit pyrosim versions, the former on a dualcore and the latter on a quadcore processor, with same results.
Somebody knows where is the problem on parallel simulations? Thanks

Michele

[Charlie Thornton]

Can you post the FDS file?  I haven''t been able to reproduce the problem on a test model, but it didn''t have quite the same configuration.  I did get a problem when I tried to adjust switchgear to be 150k x 150k:

Code:

forrtl: severe (67): input statement requires too much data, unit 21, file C:Users horntonDesktopprobsswitchgearuntitled_0002.restart

Image              PC                Routine            Line        Source             
fds5_mpi_win64.ex  000000014051DE58  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  000000014051CBA9  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404ECCC6  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404C86A1  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404C7ECF  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404A809A  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404A5FA9  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001401C8C6D  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404237E7  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  00000001404C334C  Unknown               Unknown  Unknown
fds5_mpi_win64.ex  0000000140456F35  Unknown               Unknown  Unknown
kernel32.dll       00000000777BBE3D  Unknown               Unknown  Unknown
ntdll.dll          00000000779C6A51  Unknown

But you''re talking about not seeing any messages, so it''s probably different.  My first test with switchgear (fewer cells) didn''t show any problems.  Debugging MPI can get frustrating because a lot of the output ends up being suppressed.

[Michele]

Attached the FDS file, but it is a complicated one. I did not have the time to simplify the file and find out where the problem is. Moreover I attach the "FDS simulation" window. As you see there is no error, but after more that 10 minutes there is no further time steps. The only action I can do is manually kill the program via Taskmanager.

[Charlie Thornton]

I just get a "forrtl: severe (157)" when I try to run it initially (both parallel and serial mode).  What version of FDS are you using?

[Michele]

FDS 32bit Version 5.4.3