Performance evaluation of StarPU schedulers with preconditioned conjugate gradient solver on heterogeneous (multi-CPUs/multi-GPUs) architecture

摘要

We consider the problem of scheduling sparse linear application on heterogeneous (CPUs/GPUs) platform. More specifically, we focus on the preconditioned conjugate gradient solver (PCG) since it exhibits the main features of such problems. Indeed, the relative performance of CPU and GPU highly depends on the sub-routine: GPUs are for instance much more efficient to process regular kernels such as matrix vector multiplications rather than more irregular kernels such as matrix factorization. In this context, one solution consists in relying on dynamic scheduling and resource allocation mechanisms such as the ones provided by StarPU. In this paper we evaluate the performance of dynamic schedulers proposed by StarPU, and we analyse the scalability of PCG algorithm. We show how effectively we can choose the best combination of resources in order to improve their performance.