On Fri, 7 Feb 1997, James Cownie wrote:
> Date: Fri, 07 Feb 1997 09:38:45 +0000
> From: James Cownie <jcownie@dolphinics.com>
> To: Tony Skjellum <tony@aurora.cs.msstate.edu>
> Cc: mpi-coll@mcs.anl.gov
> Subject: Re: 2-phase collective
>
>
> Tony (and Nathan) said
> > Nathan and I have been discussing this, and we will put our two cents
> > into the pot:
> >
> > 1) Why not support all three modes?
>
> but I don't understand what their proposal is.
>
> Is it
>
--> this one
> 1) 3 pairs of functions for each two faced collective
>
> start_non_collective_foo
> end_collective_foo
>
> start_collective_foo
> end_non_collective_foo
>
> start_collective_foo
> end_collective_foo
>
> or
>
> 2) Simply stating (as was my suggestion) that both the start and end
> functions are collective.
>
> Remember that the definition of collective in MPI is only that the
> call *may* require interaction with other processes, not that it
> *must* require such interaction. Therefore stating that both the
> start and end functions are collective allows any of the
> implementations, and is really just a restriction on the user code,
> forbidding things like
>
> start_foo
> send--------------->recv
> start_foo
>
> -- Jim
>
> James Cownie
> Dolphin Interconnect Solutions
> Phone : +44 117 9071438
> E-Mail: jcownie@dolphinics.com
>
>
>
Anthony Skjellum, PhD, Associate Professor of Computer Science;
Mississippi State University, Department of Computer Science & NSF ERC
Butler, Rm 300, PO Box 9637, Corner of Perry&Barr, Mississippi State,MS 39762
(601)325-8435 FAX: (601)325-8997; http://www.erc.msstate.edu/~tony;
"Persistence is fertile." ; e-mail: tony@cs.msstate.edu; Try MPI!