Problem with RESTART on different machine

[knewlands]

Hello,

I'm using code_saturne version 2.0.5 and I have encountered an error when attempting to restart a calculation previously performed on a different computer. I ran a steady state RANS on a less powerful machine and would like to use this calculation as a restart for an LES calculation on a remote computer. I have recreated the file structure on this remote machine using 'code_saturne create' and I have copied all the files across. I then created a new GUI .xml file on the remote machine and specified the restart file I'd copied across. I must've done something wrong though, as I received the following error:

SIGSEGV signal (forbidden memory area access) intercepted!

Call stack:
1: 0x35f5430280 ? (?)
2: 0x2b241c38b4ae <uiusar_+0x7d> (libsaturne.so.0)
3: 0x2b241c49fe82 <iniusi_+0x11de> (libsaturne.so.0)
4: 0x2b241c49e5a3 <initi1_+0x23> (libsaturne.so.0)
5: 0x2b241c28ab0d <cs_run+0xe1> (libsaturne.so.0)
6: 0x2b241c28b8e5 <main+0x1fa> (libsaturne.so.0)
7: 0x35f541d974 <__libc_start_main+0xf4> (libc.so.6)
8: 0x401dd9 <main+0x101> (cs_solver)
End of stack

In an attempt to understand where the problem was, I copied the .xml file from the remote machine onto my local computer and tried to run the same simulation. It works. I really need to be able to run simulations on the remote computer though, so I would greatly appreciate any help in understanding what I did wrong or what the best procedure is to successfully restart a simulation on a different machine.

Thank you,

Kristin

[Yvan Fournier]

Hello,

Does your case include mesh joining ? A small percentage of joined faces may be different due to rounding differences across machines, so the face values may be off.
If this is the case, simply do not read the restart auxiliary file (either unchecking the matching box in the advanced options for restart using the GUI, or setting ileaux = 0 in usini1.f90).
Without the auxliary file, mass fluxes are reconstructed instead of read, so you will have a slight "jump" for a few iterations, but this should solve the issue.

Regards,

Yvan