Hi Ross,
I really don't know just how thin MPI 2 support currently is, but I would
expect we may have some customers with old software on old machines. I know
it is a bit of an #ifdef pain, but if I were you I would consider
introducing this as a new default, with a conditional that will allow
compilation and functionality under mpi 1; I would do this for at least one
release, just to be nice to the guys that buy our stuff :-) Then, if mpi 2
is available pretty much everywhere, cut over to mpi 2 only in 11.
Best Regards - Bob
----- Original Message -----
From: "Ross Walker" <ross.rosswalker.co.uk>
To: <amber-developers.scripps.edu>
Sent: Monday, October 16, 2006 11:11 PM
Subject: amber-developers: MPI 2 for Amber 10?
> Hi All,
>
> For Amber 10 we will be requiring a Fortran 95 compiler. Will we also be
> requiring an implementation of MPI v2?
>
> There are some functions of MPI 2 that I would like to use in the QMMM
> routines but I don't want to have to put in ifdefs to code both MPI v1 and
> MPI v2 variants.
>
> Most new machines should support mpi v2. MPICH 2 does as does OPENMPI.
>
> Lam does not support it (although it does seem to support MPI_IN_PLACE).
> Although from the error messages generated by the compiler with the latest
> configure script it would appear that the mpif.h provided by lam is not
> stictly Fortran 95 compliant anyway.
>
> If we go down this route it means we can start using in place varients of
> most mpi calls (which is standard in mpi v2).
>
> Comments?
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | HPC Consultant and Staff Scientist |
> | San Diego Supercomputer Center |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk | PGP Key available on request |
>
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
>
>
>
Received on Wed Oct 18 2006 - 06:07:22 PDT
