Dear SIESTA users, can you help me to write request for resources for parallel run. It looks like when this request is incorrect the job stops.
We are using PBS Pro 9.2 system. The request string looks like this: #PBS -l select=3:ncpus=3:mpiprocs=3:mem=6GB The system I am studying is Si slab about 200 atoms, k-points mp(221), DZP basis set. The question is how to write resource request for a) The case of parallel over orbitals run b) The case of parallel over k-points run Are there any rules how the resources should be requested? I mean how the number of k-points or number of orbitals should relate to number of mpiprocs/ncpus/select/mem? The machine we are using is: Intel X5355 (2666 MHz) 8 Mb L2 Cache 64 servers, each is double 4 cores (512 cores total) 16 Gb memory each core Kind regards Ruslan
