Hi Sean, 2015-10-28 11:02 GMT-06:00 Hefty, Sean <[email protected]>:
> It's been a while, but we've had several discussions around the > introduction of 'strided IOVs'. To recall, the goal of a strided IOV is to > allow an application to reference every Nth element of an array using a > compact data structure, versus the more tradition struct iovec. > > In order to define this properly, I need to understand what interfaces > strided iov's make sense. For reference, see github issue 32: > > https://github.com/ofiwg/libfabric/issues/32 > > I can envision a use case for RMA, but adding this to the msg and tagged > interfaces does not seem useful. Is this correct? What about atomics? > Unfortunately, I don't think one can rule out a use case for strided iov's for tagged messages. MPI in particular supports the notion of derived data types, including srided types, which can be used for send/receive of messages. Hardware providers that have optimizations for these types of patterns would likely want to have such a strided iov interface for send/recv. > > - Sean > _______________________________________________ > ofiwg mailing list > [email protected] > http://lists.openfabrics.org/mailman/listinfo/ofiwg >
_______________________________________________ ofiwg mailing list [email protected] http://lists.openfabrics.org/mailman/listinfo/ofiwg
