End-to-End Delay Minimization in Thermally Constrained Distributed Systems

Pratyush Kumar and Lothar Thiele
ETH Zurich


With ever-increasing power densities, managing on-chip temperatures by optimizing mapping and scheduling of tasks is becoming increasingly necessary. We study the minimization of end-to-end delay for thermally constrained scheduling of an application, specified as a task graph, executing on parallel processors without speed scaling. For a given mapping of tasks to processors, we design the provably optimal schedule and present a simple implementation of such a schedule with temperature sensors. We then present different thermal-aware modifications to standard mapping heuristics and evaluate them on a large set of problem instances. The experimental results illustrate that with simple thermal-aware modifications, large improvements in end-to-end delay can be achieved.