However, the standard does contain the following (MPI-3.0 page 175) *Advice to implementors.* *It is strongly recommended that MPI_REDUCE be im-* *plemented so that the same result be obtained whenever the function is applied on* *the same arguments, appearing in the same order. Note that this may prevent op-* *timizations that take advantage of the physical location of ranks. (End of advice to* *implementors.)* It would be useful to have a table of which implementations adhere to this advice. On Mon, Nov 12, 2012 at 9:35 PM, Pavan Balaji <[email protected]> wrote:
The MPI standard does not guarantee this. For example, an implementation could use topology-aware reductions or an MPI_ANY_SOURCE while waiting for messages from peers making the order in which the reduction operations are applied different for each run.