MPI_GET_COUNT for nonblocking collective operations?
Dear MPI forum, probably it doesn't make much sense to use MPI_GET_COUNT to get a count for nonblocking collective operations (since there is only one single count for send and receive), but nevertheless IMHO there should be a comment about the behaviour of this combination in the MPI standard. Looking at the MPI Version 3.0 "green book", there's a statement that MPI_SOURCE and MPI_TAG are undefined in this context on page 197 line 32f, but there seems to be nothing about MPI_GET_COUNT here as well as on page 31 in the MPI_GET_COUNT section. Sorry, if I just missed something here. (But other persons obviously also didn't see it, e.g. http://www.open-mpi.org/community/lists/users/2013/04/21777.php) Best regards, Martin Bernreuther -- ________________________________________________________________________ Dr.-Ing. Martin Bernreuther University of Stuttgart High Performance Computing Center (HLRS) Nobelstrasse 19 (Office: 1.307) 70569 Stuttgart, Germany Phone: (++49-(0)711) 685-64542 Fax: (++49-(0)711) 685-54542, -65832 E-Mail: [email protected] URL: http://www.hlrs.de/people/bernreuther/
participants (1)
-
Martin Bernreuther