Intra-Node Memory Safe GPU Co-Scheduling

Carlos Reaño

doi:10.1109/TPDS.2017.2784428

Intra-Node Memory Safe GPU Co-Scheduling

2018-05-01

8

OA

AI

Dimitrios S. Nikolopoulos

B

Blesson Varghese *

DOI：10.1109/TPDS.2017.2784428

Original

Save

View PDF

Abstract

En 中文

GPUs in High-Performance Computing systems remain under-utilised due to the unavailability of schedulers that can safely schedule multiple applications to share the same GPU. The research reported in this paper is motivated to improve the utilisation of GPUs by proposing a framework, we refer to as schedGPU, to facilitate intra-node GPU co-scheduling such that a GPU can be safely shared among multiple applications by taking memory constraints into account. Two approaches, namely a client-server and a shared memory approach are explored. However, the shared memory approach is more suitable due to lower overheads when compared to the former approach. Four policies are proposed in schedGPU to handle applications that are waiting to access the GPU, two of which account for priorities. The feasibility of schedGPU is validated on three real-world applications. The key observation is that a performance gain is achieved. For single applications, a gain of over 10 times, as measured by GPU utilisation and GPU memory utilisation, is obtained. For workloads comprising multiple applications, a speed-up of up to 5x in the total execution time is noted. Moreover, the average GPU utilisation and average GPU memory utilisation is increased by 5 and 12 times, respectively.

Keywords:

GPU co-scheduling

access synchronisation

memory safe

accelerator

under-utilisation

schedGPU

Intra-Node Memory Safe GPU Co-Scheduling

Abstract

AI Summary

Journal

Organization