[Nek5000-users] Seg. fault in running a backward facing step case.

nek5000-users at lists.mcs.anl.gov nek5000-users at lists.mcs.anl.gov
Fri Apr 9 11:30:20 CDT 2010


For the moment you're ok if:

TORDER=2, LORDER=3, LELT=NELT/NP + 1
TORDER=2, LORDER=3, LELT= (3/2) * NELT/NP + 1
TORDER=3, LORDER=3, LELT= (4/3) * NELT/NP + 1

hth,
Stefan

On Apr 9, 2010, at 6:12 PM, nek5000-users at lists.mcs.anl.gov wrote:

> Thanks for the help!!
> 
> Harish
> 
> On Fri, 9 Apr 2010 nek5000-users at lists.mcs.anl.gov wrote:
> 
>> Yes that's true. I'll discuss with the other developers how this can be improved. Also we're working on a fix to avoid the SEGFAULT.
>> 
>> Stefan
>> 
>> 
>> On Apr 9, 2010, at 6:05 PM, nek5000-users at lists.mcs.anl.gov wrote:
>> 
>>> Hi Stefan,
>>> 
>>> Thanks for your reply. Compiling with with a larger lelt and OIFS on the code worked.
>>> 
>>> So what I understand from your reply is that if OIFS is ON the code requires more memory
>>> and it should be compiled with more lelt rather than the exact lelt=lelg/lp.
>>> 
>>> Harish.
>>> 
>>> 
>>> On Fri, 9 Apr 2010 nek5000-users at lists.mcs.anl.gov wrote:
>>> 
>>>> Harish,
>>>> 
>>>> when we try to exit the code with an error message the code crashes. You turned on OIFS (T IFCHAR) but you don't have enough memory available. I think you can just increase lelt to avoid this problem. Anyway the code should print out a reasonable err-msg and not a SEGFAULT!
>>>> 
>>>> 
>>>> Stefan
>>>> 
>>>> 
>>>> 
>>>> On Apr 9, 2010, at 4:33 PM, nek5000-users at lists.mcs.anl.gov wrote:
>>>> 
>>>>> Hello,
>>>>> 
>>>>> I am trying to simulate flow in a 3D backward facing step (BFS) geometry.
>>>>> 
>>>>> I have attached the *.box file and n2to3.in file that I use to create the
>>>>> 3D geometry. For reference logfile, *.rea and *.usr files are attached.
>>>>> The code is compiled with GNU compiler and OpenMPI. The code is compiled
>>>>> for 16 processors and compiler.out file is attached as well.
>>>>> 
>>>>> When I run the code, I get a seg. fault. After running the "gdb" tool on
>>>>> the core.* file, I get the following output. I have attached the gdb
>>>>> output file too.
>>>>> ===========================================================================
>>>>> Core was generated by `./nek5000'.
>>>>> Program terminated with signal 11, Segmentation fault.
>>>>> [New process 7652]
>>>>> #0  0x0000000000488768 in chcopy_ ()
>>>>> ===========================================================================
>>>>> 
>>>>> 3D BFS geometry has a wall at the bottom and symmetry b.c. at the top. But
>>>>> if I change the top b.c. with wall the simulation works. For reference I
>>>>> have attached the job output file.
>>>>> 
>>>>> Initially, I thought the problem was with memory limitations on my cluster
>>>>> for a bigger geometry. So I tried a smaller geometry, but I still kept
>>>>> getting seg. faults.
>>>>> 
>>>>> Help with solving this problem will be greatly appreciated. And sorry for
>>>>> attaching so many files.
>>>>> 
>>>>> Thanks,
>>>>> 
>>>>> Harish.<bfs.box><n2to3.in><logfile><bfs3d.rea><bfs3d.usr><Backstp.2323.o><gdb.output><compiler.out>_______________________________________________
>>>>> Nek5000-users mailing list
>>>>> Nek5000-users at lists.mcs.anl.gov
>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>> 
>>>> _______________________________________________
>>>> Nek5000-users mailing list
>>>> Nek5000-users at lists.mcs.anl.gov
>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>> 
>>> _______________________________________________
>>> Nek5000-users mailing list
>>> Nek5000-users at lists.mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> 
>> _______________________________________________
>> Nek5000-users mailing list
>> Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> 
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users




More information about the Nek5000-users mailing list