Efficient gather and scatter operations on graphics processors

Publisher Website
Google Scholar
Abstract
Gather and scatter are two fundamental data-parallel operations, where a large number of data items are read (gathered) from or are written (scattered) to given locations. In this paper, we study these two operations on graphics processing units (GPUs). With superior computing power and high memory bandwidth, GPUs have become a commodity multiprocessor platform for general-purpose high-performance computing. However, due to the random access nature of gather and scatter, a naive implementation of the two operations suffers from a low utilization of the memory bandwidth and consequently a long, unhidden memory latency. Additionally, the architectural details of the GPUs, in particular, the memory hierarchy design, are unclear to the programmers. Therefore, we design multi-pass gather and scatter operations to improve their data access locality, and develop a performance model to help understand and optimize these two operations. We have evaluated our algorithms in sorting, hashing, and the sparse matrix-vector multiplication in comparison with their optimized CPU counterparts. Our results show that these optimizations yield 2-4X improvement on the GPU bandwidth utilization and 30-50% improvement on the response time. Overall, our optimized GPU implementations are 2-7X faster than their optimized CPU counterparts. (c) 2007 ACM
Funding Information
  • Research Grants Council, University Grants Committee, Hong Kong (DAG05/06/EG11617206)

This publication has 14 references indexed in Scilit: