Design considerations for GPU‐aware collective communications in MPI

Abstract

Summary GPU accelerators have established themselves in the state‐of‐the‐art clusters by offering high performance and energy efficiency. In such systems, efficient inter‐process GPU communication is of paramount importance to application performance. This paper investigates various algorithms in conjunction with the latest GPU features to improve GPU collective operations. First, we propose a GPU Shared Buffer‐aware (GSB) algorithm and a Binomial Tree Based (BTB) algorithm for GPU collectives on single‐GPU nodes. We then propose a hierarchical framework for clusters with multi‐GPU nodes. By studying various combinations of algorithms, we highlight the importance of choosing the right algorithm within each level. The evaluation of our framework on MPI_Allreduce shows promising performance results for large message sizes. To address the shortcoming for small and medium messages, we present the benefit of using the Hyper‐Q feature and the MPS service in jointly using CUDA IPC and host‐staged copy types to perform multiple inter‐process communications. However, we argue that efficient designs are still required to further harness this potential. Accordingly, we propose a static and a dynamic algorithm for MPI_Allgather and MPI_Allreduce and present their effectiveness on various message sizes. Our profiling results indicate that the achieved performance is mainly rooted in overlapping different copy types.

Heterogeneous and unconventional cluster architectures and applications

“…In particular, this selection includes four interesting works . Two of them were contributions from the last two workshop editions (HUCAA 2015 and 2016, both collocated with the International Conference on Parallel Processing ‐ ICPP'15 and ICPP'16).…”

Molecular Docking for Ligand-Receptor Binding Process Based on Heterogeneous Computing

“…Usually, scientists directly use the way of MPI parallelism or GPU parallelism. However, MPI and GPU can be combined and used for large-scale computing tasks in many fields, and the MPI-GPU heterogeneous way has been widely used [29][30][31]. In computational fluid dynamics, Choi et al used a floating-point compression algorithm to optimize the GPU memory capacity in the heterogeneous MPI-GPU implementation [32], and Lai et al developed a heterogeneous parallel program combining MPI and CUDA for CFD applications on high-performance computing clusters to greatly improve computational efficiency [33].…”