Sciweavers

IEEECIT
2010
IEEE
13 years 9 months ago
Exploiting More Parallelism from Applications Having Generalized Reductions on GPU Architectures
Reduction is a common component of many applications, but can often be the limiting factor for parallelization. Previous reduction work has focused on detecting reduction idioms a...
Xiao-Long Wu, Nady Obeid, Wen-Mei Hwu
IPPS
2009
IEEE
14 years 5 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 ...
Torsten Hoefler, Jesper Larsson Träff