In the examples directory, ph5example.c shows how to use MPIIO to read/write datasets in parallel.
The function phdf5writeAll() writes 2 datasets collectively, where the first dataset is written using collective I/O and the second dataset is written using the independent mode.
On line 643:
the transfer properties for the second dataset are set. Should the last argument be H5FD_MPIO_INDEPENDENT instead of H5FD_MPIO_COLLECTIVE, since this dataset is intended to be written using the independent mode? Please correct me if I am wrong. If the example is correct, how does every process write independently?