4th Symposium on
Search Based Software Engineering
September 28th - 30th, 2012
Riva del Garda | Trento | Italy
Home » Tutorials
Mon, 25 September 2017 19:02:21


High Performance SBSE Using Commodity Graphics Cards

Speaker: Simon Poulding

Abstract: In contrast to manual software engineering techniques, search-based software engineering (SBSE) is able to exploit high performance computing resources in order to improve scalability and to solve hitherto intractable engineering problems. This is an increasingly viable approach: affordable high performance computing is readily available using cloud services, server clusters, and-perhaps surprisingly-commodity graphics cards that may already be in your laptop or PC. Modern graphics cards render high resolution, high frame-rate graphics in real-time to support the demands of applications such as gaming. To achieve this performance, the architecture of the graphics processing unit (GPU) is designed to execute a large number of threads in parallel, and to minimise the overhead of memory access. Over the last few years, this high performance parallel computing environment has become available for use by applications other than graphics rendering, a technique known as general-purpose computing on graphics processing units (GPGPU). This tutorial is an introduction to GPGPU and how it can be used by SBSE techniques. Firstly, I will demonstrate how to develop applications that run on GPUs, highlighting the important architectural differences between CPUs and GPUs, and discussing the tools that support development. Secondly, I will illustrate how the high-performance GPGPU environment may be exploited by SBSE techniques (a) to accelerate fitness evaluation, and, (b) to apply sophisticated search algorithms.

Simon Poulding is a Research Associate in the Department of Computer Science at the University of York, UK. After a successful career as a software engineer in the finance and healthcare industries, Simon returned to academia in 2005. He was a member of the multi-institution SEBASE project that made major contributions to the search-based software engineering field, and is currently a member of DAASE (Dynamic Adaptive Automated Software Engineering), a collaborative project involving four UK universities and industrial partners. In addition to search-based software engineering, Simon's research interests are software testing, general purpose computation on GPUs, model driven engineering, and the use of reliable and efficient empirical methods in computer science.

Last modified Wednesday, 03rd October, 2012