57. Power-Aware Heterogeneous Computing Through CPU-GPU Hybridization
Authors: Kyle Siehl (Washington State University Vancouver)Xinghui Zhao (Washington State University Vancouver)
Abstract: Graphic Processing Units (GPUs) have recently been widely used in general purpose computing, aiming for improving the performance of applications. However, this performance gain often comes with higher power consumption. In this paper, we present Archon, a framework for power-aware CPU-GPU hybridization. Specifically, Archon takes user's programs as input, automatically distribute the workload between CPU and GPU, and dynamically tunes the distribution ratio at runtime for an energy-efficient execution. Experiments have been carried out using a matrix multiplication application, and the results show that Archon can provide considerable energy savings, comparing to the CPU-only and GPU-only executions. These energy savings are achieved without extra effort from the programmers.
Two-page extended abstract: pdf