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

Re: T3D NetCDF implementation at LLNL/NERSC

Hi Steve,

> Tim Williams, a physicist at LLNL/NERSC, mentioned to you a couple weeks
> ago that we have a parallel version of NetCDF running on the T3D at NERSC.
> I'd like to fill you in on the changes that I made to get "libnetcdf.a"
> working on the T3D.

I think I told Tim that we'd have a new release incorporating our Cray
optimizations available by now, but it turns out I was being too
optimistic.  It now looks like it will be another two weeks.  I'm glad
you've provided him with something that sounds like it works even better
on the T3D than the Cray optimizations in netCDF 2.4.

I've taken the liberty of passing on your notes about the FFIO layer, XDR,
and your parallel I/O library to the developer of the FFIO layer here, Jeff
Kuehne (address@hidden).

> I'll append my changes and XDR and "par_io" code in uuencoded,
> compressed, tar format.

Thanks, it unpacked with no problems.  We won't be able to integrate this
into netCDF 2.4, but I'd like to make your changes available to anyone who
asks about netCDF on the T3D, and to refer them to you if they have
questions.  I hope you'll get a chance to test if your changes work OK with
the netCDF 2.4 release, when we finally announce it.  NCAR recently got
access to a T3D, so we could theoretically test and maintain this locally
ourselves, but we'll need to learn a bit more about it first.

Thanks again for making your changes available.

______________________________________________________________________________

Russ Rew                                           UCAR Unidata Program
address@hidden                              http://www.unidata.ucar.edu