The semantics of nonblocking partitioned communication are defined by suitably extending the definitions in Section Semantics of Point-to-Point Communication.
Interpretation of count and datatype for partitioned communication. Partitioned communication uses the count and datatype arguments in the partitioned communication initialization functions to describe a single partition. The argument partitions specifies how many equal partitions of a number ( count) of objects of datatypes make up the entire buffer to be transferred in the partitioned communication. As partitioned communication describes many partitions, using absolute displacements in datatypes (e.g., MPI_BOTTOM) is not supported. Partitions are contiguous in memory, there is no padding in between them. Once a partitioned send operation is started, each partition must be marked as ready using MPI_PREADY and the operation must be completed using a completion function, such as MPI_TEST or MPI_WAIT.
Order. Matching follows the same MPI matching rules as for point-to-point communication (see Chapter Point-to-Point Communication) with communicator, tag, and source dictating message matching. In the event that the communicator, tag, and source do not uniquely identify the message, the order in which partitioned communication initialization calls are made is the order in which they will eventually match.