An optimizing data decomposition method based on linear inequalities
Zheng Wang · Journal of Computer Applications · 2007
Maximize parallelism and minimizing communication by increasing the locality of data references are important issues for achieving high performance on large-scale parallel machines.This paper presented an optimizing method for data calculation and computation decomposition.The authors have improved the method by Anderson and Lam.Use linear inequality system to describe data and computation decomposition and reconstruct linear inequality system after adding read-only array with boundary replication.Communication load can be reduced effectively.