Embedded tutorial — Compilation techniques for CGRAs: Exploring all parallelization approaches

Abstract

Coarse-Grained Reconfigurable Array (CGRA) processors accelerate inner loops of applications by exploiting instructionlevel parallelism (ILP) and in some cases also data-level and task-level parallelism (DLP & TLP). The aim of this tutorial is to give insight in CGRA architectures and their compilation techniques to exploit parallelism. These topics will be covered: · Polymorphic pipeline arrays, expanding coarse-grained arrays beyond innermost loops (Scott Mahlke, University of Michigan) · Code-generation for coarse-grained arrays: flexibility and programmer productivity (Bjorn De Sutter, Ghent University) · Memory-aware compilation techniques for CGRAs (Aviral Shrivastava, Arizona State University) · Retargetable Mapping of Loop Programs on Coarse-grained Reconfigurable Arrays (Frank Hannig, University of Erlangen-Nuremberg).