[gmx-users] why the trajectory file is not output

[chris.neale at utoronto.ca]

Hello Yang,

It was asked twice if you ran serial or parallel and I did not see an  
answer to that question.

we have seen segmentation faults with 3.3.1 and openmpi when running  
complex systems in parallel for systems that ran fine in serial. It  
was certainly a head-scratcher as it did seem to depend on having many  
different types of molecules in the same box. The solution for us at  
that time was to use lam-mpi, but only for systems that crashed using  
openmpi since lam-mpi is slower. I also note that this problem has  
disapeared in gmx-4 for us.

Chris.

-- original message --

I just choose a small number of steps but it also does not work.it  
always shows that

starting mdrun 'MODEL CYLINDER'
10000 steps,    300.0 ps.

Segmentation fault (core dumped)

I am not very sure about that.

Thank you .

Yang