[Mpi3-rma] Updated Proposal 1

Torsten Hoefler htor at illinois.edu
Wed Dec 8 10:28:34 CST 2010

On Wed, Dec 08, 2010 at 12:22:45AM -0600, Pavan Balaji wrote:
> On 12/07/2010 11:32 PM, Torsten Hoefler wrote:
>> 1) moved the lockfree (or conflictfree) mode to the JoD
> Sorry, Torsten :-).
No worries at all :-). It was only a naming issue and I think the current
name is horrible and disgusting. But we would have to have that anyway
and adding a second name for makeup was also suboptimal ;-).

>> 2) added a discussion about ordering (Section 11.8.2)
> The new discussion is better, but:
> (a) I think the ordering discussion requires some edits to some places  
> earlier in the chapter too (see proposal 2).
Yes, there are some places that now need forward-references (I didn't do
the complete cleanup yesterday because it was late, but I have a note).

> (b) Do we know any network (except ones that emulate one-sided  
> communication over two-sided messaging) that provide strict ordering? As  
> far as I know, most networks only provide ordering between PUTs (or  
> atomic PUTs). Ordering between GETs is also not provided, unless the  
> user does a flush after each GET operation. Is this only for convenience?
I don't know. Just that you're aware: the current default would force
the MPI implementation to flush after each get. However, I am also
planning to add an advice to users somewhere *very visible* that the
user should really strive to provide these info arguments if possible
(because the performance difference could be hideous). So the editing is
not finished but 11.8.2 is ready for discussion ;-).


 bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ -----
Torsten Hoefler         | Performance Modeling and Simulation Lead
Blue Waters Directorate | University of Illinois (UIUC)
1205 W Clark Street     | Urbana, IL, 61801
NCSA Building           | +01 (217) 244-7736

More information about the mpiwg-rma mailing list