ParCo 2009: MS on GPU Computing

EuroGPU 2009 -- Introduction and Overview

The success of the gaming industry is now pushing processor technology like we have never seen before. Since recent graphics processors (GPU's) have been improving both their programmability as well as have been adding more and more floating point processing, it makes them very appealing as accelerators for general-purpose computing. This minisymposium gives an overview of some of these advancements by bringing together experts working on the development of techniques and tools that improve the programmability of GPU's as well as the experts interested in utilizing the computational power of GPU' scientific applications.

This first EuroGPU Minisymposium held as a minisymposium at ParCo 2009 in Lyon, France on Sept. 1-2, 2009, brought together severl experts working on the development of techniques and tools that improve the programmability of GPU's as well as the experts interested in utilizing the computational power of GPU's for scientific applications. This short summary thus gives a very useful, but quick overview of some of the major recent advancement in modern GPU computing.

The minisymposium started with a short history and overview as seen by Anne C. Elster, one of the organizers. And introduction to GPU computing and EuroGPU09 entitled "Parallel Computing with GPUs" by orgainzers Anne C. Elster and Stephane Requena was included in the IOP Proceedings. (PDF for personal use)

The introductory talk went directly into her work with her graduate student Rune J. Hovland on Thoughput Computing on Future GPUs, (PDF for personal use) .

The rest of the minisymposium was divided into three parts:

1. an industry track
2. an academic track with 4 papers that will be included in the Parco 2009 Proceedings published by IOS Press (Editors: Joubert et al.) and
3. a panel discussion.

Panel discussion on GPU Computing

The panel discussion started with an intro by Rune J. Jensen (CERN/NTNU) on Compiler Issues and Challenges where he pointed out how different compiler versions of gcc and icc can give different performance results for a given block-size and may even skip calculations. He was able to find these issues though the use of lots of performance counters, which led to the questions of when we can expect such HW counters on GPUs.

The panel was organized by Anne C. Elster and included the following participants:

• Guillamme Colin de Verdriere, CEA, France
• Anne C. Elster, Norwegian Univ. of Science and Technology (NTNU), Norway
• Rune J. Jensen, CERN, Switzerland/NTNU Norway
• Tim Lanfear, NVIDIA, UK
• Stephane Vialle, Supelec Metz, France

and lead to interesting discussions with views from both academia and industry.

ParCo2009 Conference Proceedings

The printed version of the conference proceedings, including the academic papers presented in this Minisyposium, is available. Please check http://www.iospress.nl.

The title of the book is:

Parallel Computing: From Multicores and GPU's to Petascale

Volume 19 Advances in Parallel Computing
Edited by: B. Chapman, F. Desprez, G.R. Joubert, A. Lichnewsky, F. Peters and T. Priol
April 2010, 760 pp., hardcover
ISBN: 978-1-60750-529-7

Electronic Version of the ParCo2009 Proceedings

An electronic version of the proceedings will be available soon at http://www.iospress.nl then select Book Series and Advances in Parallel Computing. Then click "Go to e-Book". Individual copies of papers can be ordered. These copies contain colour graphics and images in contrast to the printed version. Aternatively go to  http://www.booksonline.iospress.com.