Hi, Thanks Lex and Lorentzo for your answers. I cannot tell for sure that the cluster is set up right. It is a recent cluster that is still in testing phase, and which I do not maintain. I would be grateful if you could please give me some indication on how to help the cluster manager locate the problem and solve it... Please just let me know if you need information from me. I used the quick Q-E compilation, which seemed to occur normally. Maybe I could find an installation logfile to help (just don't know where to find it).
What makes me think the problem comes more form Q-E is that the job is well distributed on the processors of all nodes as soon as the Q-E (right?) option "-in inputfile.in" is used (as explained in Q-E user guide, see link below) instead of the standard "pw.x <inputfile.in> outputfile.out" syntax. http://www.quantum-espresso.org/wiki/index.php/Running_on_parallel_machines#Trouble_with_input_files If as Lorentzo wrote it is not possible to redirect the output, then it is sad, since I found it particularly convenient to be able to run series of calculations using a single script, as is typically done in the example files. This is particularly convenient when looking for the best cutoff energy, for example. Best regards. Sylvian. -- ------------------------------------------------------------------------ *Sylvian CADARS* CEMHTI-CNRS 1D, Av. de la Recherche Scientifique 45071 Orl?ans Cedex 2, France ------------------------------------------------------------------------
