PIO is a software interface layer designed to encapsolate the complexities of parallel IO and make it easier to replace the lower level software backend. It currently supports netcdf (http://www.unidata.com), pnetcdf (http://trac.mcs.anl.gov/projects/parallel-netcdf) and MPI-IO.
PIO calls are collective. A MPI communicator is set in a call to PIO_init and all tasks associated with that communicator must participate in all subsequent calls to PIO. An application can make multiple calls to PIO in order to support multiple MPI communicators.
To use PIO, your program should begin by calling the PIO_init function providing the MPI communicator and the rank within that communicator of the calling task. You should also provide the number of iotasks to be used, the stride or number of tasks between iotasks, and the number of mpi aggregators to be used. You may optionally also choose the base IO task. This task will be used for output of any non-decomposed data. This call initializes an IO system type structure that will be used in subsequent file and decomposition functions.
You can then open a file for reading or writing with a call to PIO_createfile or PIO_openfile. In this call you will specify the file type: pio_iotype_netcdf, pio_iotype_pnetcdf, or pio_iotype_binary, pio_iotype_netcdf4c or pio_iotype_netcdf4p; along with the file name and optionally the netcdf mode.
To read or write decomposed data you must first describe the mapping between the organization of data in the file and that same data in the application space. This is done in a call to PIO_initdecomp. In the simplest call to this function, a one dimensional integer array is passed from each task, the values in the array represent the 0 based offset from the beginning of the array on file.