Microarchitecture support for dynamic scheduling of acyclic task graphs

Abstract

Any program can be broken into its loop structure, plus acyclic dependence graphs representing the body of each loop or subroutine. The parallelism inherent in these acyclic graphs augments the loop- level parallelism available in the program. This paper presents two algorithms for dynamic scheduling of acy- clic task graphs containing both data and control dependences, and describes a microarchitecture which implements these algorithms efficiently.