[Mpi3-rma] Fix for win_allocate_shared

Barrett, Brian W bwbarre at sandia.gov
Fri Sep 23 12:47:34 CDT 2011

It might not be able to. If non-coherent load/store is byte access (or word access), it can provide a separate public and private window in separate and meet the semantics.  If it provides cacheline access to non-coherent space, it will not be able to provide shared memory across uncached space; such is the cost of standardization.


----- Original Message -----
From: Pavan Balaji [mailto:balaji at mcs.anl.gov]
Sent: Friday, September 23, 2011 11:40 AM
To: Torsten Hoefler <htor at illinois.edu>
Cc: mpi3-rma at lists.mpi-forum.org <mpi3-rma at lists.mpi-forum.org>; mpi3-hybridpm at lists.mpi-forum.org <mpi3-hybridpm at lists.mpi-forum.org>
Subject: Re: [Mpi3-rma] Fix for win_allocate_shared

On 09/23/2011 12:39 PM, Torsten Hoefler wrote:
> It clarifies semantics for separate (the first part) and the advice to
> implementers makes the semantic restriction explicit.

What I mean is -- how will an MPI implementation on non-cache-coherent 
system be able to expose shared memory with this restriction?

  -- Pavan

Pavan Balaji
mpi3-rma mailing list
mpi3-rma at lists.mpi-forum.org

More information about the mpiwg-rma mailing list