code_saturne User's Forum

Hello Yvan,

it is really strange but today i recieved again the same error as yesterday "forbidden memory area access" when running on 2,3 or 4 processors. Running in single mode works but takes a long time, of course. I do not know why it works sometimes and sometimes not. I attached the corresponding files.

Regards,

Sebastian

Hello again ,
finally i obtain rotation and it will be also vizualised in paraview. Thank you very much Yvan. But i haveobtain an error in single mode (parallel does not work as written in the previous post) that says the following:

SIGFPE signal (floating point exception) intercepted!

Call stack:
1: 0x7715c4 <> (cs_solver)
2: 0x834fa5 <> (cs_solver)
3: 0x82fec7 <> (cs_solver)
4: 0x6e2d1a <> (cs_solver)
5: 0x430677 <> (cs_solver)
6: 0x431356 <> (cs_solver)
7: 0x6d3b7c <> (cs_solver)
8: 0x45fdbe <> (cs_solver)
9: 0x438abb <> (cs_solver)
10: 0x43562d <> (cs_solver)
11: 0x413739 <> (cs_solver)
12: 0x40f0d5 <> (cs_solver)
13: 0x405e63 <> (cs_solver)
14: 0x7f122bb9176d <__libc_start_main+0xed> (libc.so.6)
15: 0x405ea5 <> (cs_solver)
End of stack

I attached the files.

Regards,

Sebastian

Hello,

Could you send me your new mesh (by private message if you prefer) ? I'll take a look at it.

Regards,

Yvan

Hello,

I fixed the bug leading to the parallel crash (which was much simpler than the joining one: a ghost cell exchange for non-symmetric tensors was used instead of one for symmetric tensors, so out-of-bounds array accesses occurred.

Until we release 3.2.2, you can add the attached file to you user subroutines.

I have a test running on 4 cores, which has not yet reached the "floating point exception" (numerical) crash, but the CFL at the first time step is still too large, so you'll probably need to reduce the time step by a factor of 20 again (the finer the mesh, the smaller a time step you need).

Regards,

Yvan

Hello,

great! Thank you very much! Do i have to use both cs_user_join.c and the new one?

Regards,

Sebastian

Another Question: in my listing file i cannot see CFL. Is it the same as Courant/Fourier Criterion? This i can find in the listing close to the end of an iteration.

Regards,

Sebastian

Hello Yvan,

with a time step of 0.0001 it was successful for 10 iterations. Could you maybe have a look to the listing? What do you think about the CFL range?

Regards,

Sebastian

Sorry, here is the listing.

Hello Yvan,

i have again a problem. I obtain this error and i do not know what to do:

At line 76 of file /home/sebastian/opt/installer/code_saturne-3.2.1/src/base/csprnt.f90 (unit = 9, file = '/dev/null')
Fortran runtime error: Insufficient data descriptors in format after reversion

=====================================================================================
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= EXIT CODE: 134
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
=====================================================================================
APPLICATION TERMINATED WITH THE EXIT STRING: Aborted (signal 6)

****************************
Saving calculation results
****************************

solver script exited with status 134.

Error running the calculation.

Check Code_Saturne log (listing) and error* files for details.

Error in calculation stage

Regards,

Sebastian

Hello Yvan,

sorry, but I have again another strange problem. My computations seems to hang at a certain time step number. Sometimes at 6, sometimes at 13 or 17. But i obtain no error. Even not after a couple of hours and all 4 processors are running on 100%. Do you have an idea what could be the reason for this?

Thank you! Regards,

Sebastian