I have been experimenting with some simple fortran test programs to write files with some of the MPI-IO functions, and have come across a troubling issue. I use the function MPI_FILE_SET_VIEW with the 'native' data representation and correctly write a file with MPI_FILE_WRITE_ALL. However, if I change to the 'external32' representation, the file is truncated, with a length that suggest that either only one process did a write, or each process wrote to the same location. I was assuming that changing to the 'external32' representation would only do a byteswap similar to what some compilers options can do with conventional fortran IO. Am I mis-interpreting the way to implement this feature?
I have attached my script/test program. T. Rosmond
write_test
Description: C-Shell script