An Adaptive Load Balancer For Graph Analytical Applications on GPUs

11/20/2019

∙

by   Vishwesh Jatala, et al.

∙

0

∙

share

Load balancing graph analytics workloads on GPUs is difficult because of the irregular nature of graph applications and the high variability in vertex degrees, particularly in power-law graphs. In this paper, we describe a novel load balancing scheme that aims to address this problem. Our scheme is implemented in the IrGL compiler to allow users to generate efficient load-balanced code for GPUs from high-level sequential programs. We evaluated several IrGL-generated load-balanced programs on up to 16 GPUs. Our experiments show that this scheme can achieve an average speed-up of 1.5x on inputs that suffer from severe load imbalance problems when previous state-of-the-art load-balancing schemes are used.