DocumentCode
2028004
Title
Enabling a Quantum Monte Carlo application for the DEEP architecture
Author
Emerson, Andrew ; Affinito, Fabio
Author_Institution
Supercomput. Applic. & Innovation (SCAI), Cineca, Casalecchio di Reno, Italy
fYear
2015
fDate
20-24 July 2015
Firstpage
453
Lastpage
457
Abstract
In the DEEP project a prototype Exascale system consisting of a standard Intel Xeon cluster linked to a “Booster” part containing Intel Xeon Phi nodes connected in a high-speed network, is being designed and constructed. In order to evaluate this novel architecture, expected to be available in the second half of 2015, a number of grand challenge applications in computational science and engineering are being modified and optimised. In this study we report on the efforts made by the Cineca project partner and DEEP support staff to enable one of these applications, the TurboRVB Quantum Monte Carlo simulation program, which can be used to study complex phenomena in materials such as superconductivity. The modified code, based on an implementation of the OmpSs offload task model, has been successfully tested on the MareNostrum supercomputer at the Barcelona Supercomputing Center.
Keywords
Monte Carlo methods; parallel machines; quantum computing; Barcelona Supercomputing Center; Cineca project; DEEP architecture; Exascale system; Intel Xeon Phi nodes; Intel Xeon cluster; MareNostrum supercomputer; OmpSs offload task model; TurboRVB Quantum Monte Carlo simulation program; high-speed network; Computer architecture; Lattices; Microwave integrated circuits; Monte Carlo methods; Programming; Prototypes; Software; Exascale Systems; Modeling and Simulation using HPC Systems; Parallelization of Simulation;
fLanguage
English
Publisher
ieee
Conference_Titel
High Performance Computing & Simulation (HPCS), 2015 International Conference on
Conference_Location
Amsterdam
Print_ISBN
978-1-4673-7812-3
Type
conf
DOI
10.1109/HPCSim.2015.7237075
Filename
7237075
Link To Document