Literature Survey on GPU Enabled Libraries for Improving Efficiency of OpenSEES on Linux Cluster

Author(s):

Shweta Nayak , Pune Institute of Computer technology, Pune; Mukta Takalikar, Pune Institute of Computer technology, Pune; Gouri Kadam, CDAC, Pune

Keywords:

GPU, OpenSEES, Cusp, CUDA, High Performance Computing

Abstract:

In recent years performance of throughput oriented applications can be considerably improved using GPU computing. It provides an alternative to CPUs because of number of cores embedded on same chip. It is observed that GPU gives performance improvement up to 10x to 100x over the CPU computing. With GPU architectural advances in improved programming support this trend will continue in the near future. Dedicated libraries such as CUDA provide general purpose computing support for GPU. SIMD architecture of GPU enabled libraries such as CUDA makes it possible to handle a very large number of hardware threads concurrently, which is beneficial for various applications, including scientific computing. This paper aims at finding the effective ways for improving the performance of OpenSEES, which is an open source package for earthquake simulation on hybrid platform using GPU enabled libraries such as Cusp.

Other Details:

Manuscript Id	:	IJSTEV2I12165
Published in	:	Volume : 2, Issue : 12
Publication Date	:	01/07/2016
Page(s)	:	322-324

Download Article