[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #VLE-656568]: Documentation for parallel I/O with f90

> Hi Ed,
>
> Thank you for making this documentation available.  You
> may ignore (close) my other ticket.  I solved all the problems
> with f77 and f90 implementations.  I already release
> a new version of ROMS with parallel I/O.  ROMS has
> very complex I/O with multiple NetCDF files for
> input and output.
>

That's wonderful.

Can you tell me more about the parallel ROMS? Do you work with the ROMS model a
lot? Where will you be releasing it?

How did the parallelization go? We you able to achieve much speedup?

Would you mind sending me your parallelized ROMS so that I can benchmark it?
ROMS has been used before for this purpose with parallel HDF5, so it would be
interesting to see a comparison.


> The only issue remaining is that the NF90_COLLECTIVE
> and NF90_INDEPENDENT is not referenced in netcdf.mod.

Thanks for pointing this out. I have added them and they are in the latest
snapshot release:
ftp://ftp.unidata.ucar.edu/pub/netcdf/snapshot/netcdf-4-daily.tar.gz

I have also added a F77 and a F90 test program for the parallel functions. See
the snapshot documentation for more info and examples.

http://www.unidata.ucar.edu/software/netcdf/docs_snapshot/netcdf-f90.html

>
> I found out that this like C logic statement where I
> can use access=0 for independent and any number
> (positive or negative; integer or floating point)
> for collective.

Yes, that is correct.

Thanks!

Ed

Ticket Details
===================
Ticket ID: VLE-656568
Department: Support netCDF
Priority: High
Status: Closed