The PWneb code is interfaced to PWscf, which is used as computational engine for total energies and forces. It can therefore take advantage from the two parallelization paradigms currently implemented in QUANTUM ESPRESSO, namely Message Passing Interface (MPI) and OpenMP threads, and exploit all PWscf-specific parallelization options. For a detailed information about parallelization in QUANTUM ESPRESSO, please refer to the general documentation.
In addition, PWneb makes several independent evaluations of energy and forces at each step of the path optimization: one per ``image'', that is, a point in the path, corresponding to a set of atomic positions. It is thus possible and often convenient to distribute images among processors, using the ``image'' parallelization, as described in the general documentation. The number of image groups is specified using the option -ni N (or, equivalently, -nimage N) after the executable name (e.g., neb.x) in the command line. The default is a single image group (no image parallelization)
Images are loosely coupled calculations: processors belonging to different image groups communicate only once in a while, whereas processors within the same image group are tightly coupled and communications are more significant (please refer to the user guide of PWscf).