Sparse collective operations for MPI

13 years 11 months ago
Sparse collective operations for MPI
Abstract—We discuss issues in designing sparse (nearest neighbor) collective operations for communication and reduction operations in small neighborhoods for the Message Passing Interface (MPI). We propose three such operations, namely a sparse gather operation, a sparse all-to-all, and a sparse reduction operation in both regular and irregular (vector) variants. By two simple experiments we show a) that a collective handle for message scheduling and communication optimization is necessary for any such interface, b) that the possibly different amount of communication between neighbors need to be taken into account by the optimization, and c) illustrate the improvements that are possible by schedules that posses global information compared to implementations that can rely on only local information. We discuss different forms the interface and optimization handles could take. The paper is inspired by current discussion in the MPI Forum.
Torsten Hoefler, Jesper Larsson Träff
Added 24 May 2010
Updated 24 May 2010
Type Conference
Year 2009
Where IPPS
Authors Torsten Hoefler, Jesper Larsson Träff
Comments (0)