[MPI3 Fortran] Sample implementation

Supalov, Alexander alexander.supalov at intel.com
Tue Apr 22 11:08:25 CDT 2008


Dear Jeff,

You're right, attribute management may be a little hard on the initial
prototype. The difference of the handle representation will best be
addressed by trying the prototype against MPICH2, etc. We can take care
of a part of that if you wish, once the prototype becomes available.

Best regards.

Alexander

-----Original Message-----
From: mpi3-fortran-bounces at lists.mpi-forum.org
[mailto:mpi3-fortran-bounces at lists.mpi-forum.org] On Behalf Of Jeff
Squyres
Sent: Tuesday, April 22, 2008 6:02 PM
To: MPI-3 Fortran working group
Subject: Re: [MPI3 Fortran] Sample implementation

On Apr 22, 2008, at 10:33 AM, Supalov, Alexander wrote:

> Thanks. I'd suggest adding some attribute copy/delete functions,
>

Ah, I didn't initially put those in the list because I put in  
ERRHANDLER functions (i.e., function pointer arguments) -- but you're  
pointing out that we should also do the whole extraordinarily- 
confusing attribute conversion stuff, right?

Should we open the possibility of treating F2003 attributes exactly  
like C attributes?  This would make the attribute situation far less  
confusing than it is now, IMNSHO...

> MPI_Request that is sometimes handled differently than other handles
> (you'll have them anyway with MPI_ISEND I guess), and MPI_WTIME, as a
> "real" function.
>

MPI_WTIME -- good call; let's add that to the list.

> MPI_File is a good example of a handle that is a pointer rather than  
> an
> integer in MPICH2, but that's probably too far away from the scope of
> this prototype.
>

Ah, I wasn't aware of that.  In OMPI, all of our handles are pointers,  
so I think if we cover a few types (eg., communicator and datatype) in  
the sample with OMPI and some other MPI that uses integers, we'll be  
covered.

So here's a new list:

6-function MPI
- MPI_INIT
- MPI_FINALIZE
- MPI_SEND
- MPI_RECV
- MPI_COMM_RANK
- MPI_COMM_SIZE
Some non-blocking functions
- MPI_ISEND
- MPI_IRECV
- MPI_WAITALL
A function with a string
- MPI_GET_PROCESSOR_NAME
Memory pointers
- MPI_ALLOC_MEM
- MPI_FREE_MEM
Varargs
- MPI_PCONTROL
Functions dealing with function pointers
- MPI_COMM_ERRHANDLER_SET
- MPI_COMM_ERRHANDLER_GET

Returns double

- MPI_WTIME

Along with at least the following pre-defined MPI handles (others can
be included of course):

Basic handles
- MPI_INTEGER
- MPI_COMM_WORLD
"Weird" handles
- MPI_BOTTOM
- MPI_STATUSES_IGNORE
Function pointer handle
- MPI_ERRORS_ARE_FATAL



-- 
Jeff Squyres
Cisco Systems

_______________________________________________
mpi3-fortran mailing list
mpi3-fortran at lists.mpi-forum.org
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-fortran
---------------------------------------------------------------------
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer
Registergericht: Muenchen HRB 47456 Ust.-IdNr.
VAT Registration No.: DE129385895
Citibank Frankfurt (BLZ 502 109 00) 600119052

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.





More information about the mpiwg-fortran mailing list