A Case for Migrating Execution for Irregular Applications
Workshop: IA^3 2017 - 7th Workshop on Irregular Applications: Architectures and Algorithms
Abstract: Modern supercomputers have millions of cores, each capable of ex- ecuting one or more threads of program execution. In these com- puters the site of execution for program threads rarely, if ever, changes from the node in which they were born. This paper dis- cusses the advantages that may accrue when thread states migrate freely from node to node, especially when migration is managed by hardware without requiring software intervention. Emphasis is on supporting the growing classes of algorithms where there is signi cant sparsity, irregularity, and lack of locality in the memory reference patterns. Evidence is drawn from reformulation of sev- eral kernels into a migrating thread context approximating that of an emerging architecture with such capabilities.