Efficient Computational Workload Distribution on Heterogeneous GPUs

Abstract

Recently, heterogeneous system architectures are becoming a mainstream for achieving high performance and power efficiency. In particular, many-core graphics processing units (GPUs) have started to play an important role for computing in heterogeneous architectures. However, for application designers, computational workload still needs to be distributed among heterogeneous GPUs manually and remains inefficient. In this work, we propose a MINLP-based method for efficient workload distribution among GPUs by considering the capabilities of GPUs for various applications. Experimental results demonstrate the performance of our proposed method.