Lattice QCD on GPU Architectures

Acronym: 
Lattice QCD on GPU Architectures (project completed)
Funding scheme: 
IRC
Funding period: 
2 years
Starting date: 
Autumn 2008
Coordinator: 
CyI/CaSToRC
Role of CyI: 
General Management and Coordination; all research aspects

Synopsis

The U(1) problem is an outstanding issue in hadron spectroscopy.  The large mass of the η′ meson, relative to the π meson, is believed to be related to QCD vacuum structure since this singlet pseudoscalar may receive a mass (in the chiral limit) due to the vacuum contribution of disconnected diagrams.

The aim of this project is to implement state-of-the-art Lattice QCD disconnected diagram techniques on graphical processing units (GPUs), as well as perform an equal-cost evaluation of these techniques. GPU architectures can yield to 130 times the processing power of conventional CPUs for scientific applications and they are well suited to Lattice QCD applications. The methodology to achieve this aim will include the development of a Lattice QCD software suite implemented on the NVIDIA GPU architecture and the selection and testing of an optimal disconnected diagram technique.

Contact person

Dr. Alan O'Cais, email: a [dot] ocais [at] cyi [dot] ac [dot] cy
Tel. +357 22208 ext. 633

The Project is co-financed by the European Regional Development Fund and the Republic of Cyprus through the Research Promotion Foundation.