[Mpi3-rma] [EXTERNAL] Re: MPI-3 UNIFIED model clarification
Barrett, Brian W
bwbarre at sandia.gov
Sun Aug 4 18:33:40 CDT 2013
On 8/4/13 5:16 PM, "Pavan Balaji" <balaji at mcs.anl.gov<mailto:balaji at mcs.anl.gov>> wrote:
On 08/04/2013 06:10 PM, Barrett, Brian W wrote:
I think we have to agree to disagree on that one, in that I really don't
think your example is a it's a common usage model worth destroying
UNIFIED over. Most one-sided codes I've seen use one-sided ops for
everything in the window, but I haven't seen as many codes as I might like.
FWIW, one case where this became a problem was when we were using
PUT/GET accesses were issued from window memory to window memory. Even
for PUT/GET accesses, the origin accesses are considered load/store
accesses; only the target accesses are considered RMA accesses.
I'm either parsing your first or second sentence incorrectly. Are you saying that you're doing a get in which the initiator-side address is in a window and that reading the data after it's arrived is a load and therefore causes problems for separate? Or are you saying that you believe that puts/gets which move memory within a window are considered loads?
I can see a use case for better supporting the first one. The second I would believe is a poor interpretation of the spec.
Brian W. Barrett
Scalable System Software Group
Sandia National Laboratories
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mpiwg-rma