number of aggregators obsolete?

[rjdave]

I am testing PIO performance inside the ROMS ocean model. In trying to figure out whether or if I should try using MPI aggregators I found the following in examples/basic/README.testpio

    rearr          - string, type of rearranging to be done
                     ("none","mct","box","boxauto")
    nprocsIO       - integer, number of IO processors used only when rearr is
                     not "none", if rearr is "none", then the IO decomposition
                     will be the computational decomposition
    base           - integer, base pe associated with nprocIO striding
    stride         - integer, the stride of io pes across the global pe set
    num_aggregator - integer, mpi-io number of aggregators, only used if no
                     pio rearranging is done

and later:

  - if rearr is set to "none", then the computational decomposition is also
    going to be used as the IO decomposition.  The computation decomposition
    must therefore be suited to the underlying I/O methods.
  - if rearr is set to "box", then pio is going to generate an internal
    IO decomposition automatically and pio will rearrange to that decomp.
  - num_aggregator is used with mpi-io and no pio rearranging.  mpi-io is only
    used with binary data.

However, at least in Fortran, there does not seem to be away to set the rearranger to "none"; only "box" and "subset" are defined in src/flib/pio_types.F90 and if I guess and try "0" (or anything but "1" or "2") PIO complains:

Abort with message unknown rearranger in file ../../../../pio/src/clib/pio_darray.c at line 921

Upon searching in this issue tracker I discovered #984 and now I'm thinking that this is just documentation left over from PIO1. Also, most of the references to the number of aggregators are in the Fortran tests so maybe those tests weren't scrutinized as much as the C tests.

In any event, am I correct in assuming I shouldn't bother attempting to test aggregators and/or "none" for rearrangement?

[jedwards4b]

Yes, num_aggregator is obsolete and rearranger should be PIO_REARR_SUBSET or PIO_REARR_BOX. Sorry for the confusion.