Run-time techniques for parallelizing sparse matrix problems

Abstract

Sparse matrix problems are difficult to parallelize efficiently on message-passing machines, since they access data through multiple levels of indirection. Inspector/executor strategies, which are typically used to parallelize such problems impose significant preprocessing overheads. This paper describes the runtime support required by new compilation techniques for sparse matrices and evaluates their performance, highlighting optimizations and improvements over previous techniques.