We present an OpenCL-based Lattice QCD application using a heatbath algorithm
for the pure gauge case and Wilson fermions in the twisted mass formulation.
The implementation is platform independent and can be used on AMD or NVIDIA
GPUs, as well as on classical CPUs. On the AMD Radeon HD 5870 our double
precision dslash implementation performs at 60 GFLOPS over a wide range of
lattice sizes. The hybrid Monte-Carlo presented reaches a speedup of four over
the reference code running on a server CPU.
%0 Journal Article
%1 Bach2012Lattice
%A Bach, Matthias
%A Lindenstruth, Volker
%A Philipsen, Owe
%A Pinke, Christopher
%D 2012
%J Computer Physics Communications
%K gpu, hpc
%N 9
%P 2042--2052
%R 10.1016/j.cpc.2013.03.020
%T Lattice QCD based on OpenCL
%U http://dx.doi.org/10.1016/j.cpc.2013.03.020
%V 184
%X We present an OpenCL-based Lattice QCD application using a heatbath algorithm
for the pure gauge case and Wilson fermions in the twisted mass formulation.
The implementation is platform independent and can be used on AMD or NVIDIA
GPUs, as well as on classical CPUs. On the AMD Radeon HD 5870 our double
precision dslash implementation performs at 60 GFLOPS over a wide range of
lattice sizes. The hybrid Monte-Carlo presented reaches a speedup of four over
the reference code running on a server CPU.
@article{Bach2012Lattice,
abstract = {{We present an OpenCL-based Lattice QCD application using a heatbath algorithm
for the pure gauge case and Wilson fermions in the twisted mass formulation.
The implementation is platform independent and can be used on AMD or NVIDIA
GPUs, as well as on classical CPUs. On the AMD Radeon HD 5870 our double
precision dslash implementation performs at 60 GFLOPS over a wide range of
lattice sizes. The hybrid Monte-Carlo presented reaches a speedup of four over
the reference code running on a server CPU.}},
added-at = {2019-02-23T22:09:48.000+0100},
archiveprefix = {arXiv},
author = {Bach, Matthias and Lindenstruth, Volker and Philipsen, Owe and Pinke, Christopher},
biburl = {https://www.bibsonomy.org/bibtex/29c070a349409806fb55ddd614adb30bf/cmcneile},
citeulike-article-id = {13600514},
citeulike-linkout-0 = {http://arxiv.org/abs/1209.5942},
citeulike-linkout-1 = {http://arxiv.org/pdf/1209.5942},
citeulike-linkout-2 = {http://dx.doi.org/10.1016/j.cpc.2013.03.020},
day = 26,
doi = {10.1016/j.cpc.2013.03.020},
eprint = {1209.5942},
interhash = {c41a6879a2286d2f4581724d0a6a6148},
intrahash = {9c070a349409806fb55ddd614adb30bf},
issn = {00104655},
journal = {Computer Physics Communications},
keywords = {gpu, hpc},
month = sep,
number = 9,
pages = {2042--2052},
posted-at = {2015-05-01 17:19:13},
priority = {2},
timestamp = {2019-02-23T22:15:27.000+0100},
title = {{Lattice QCD based on OpenCL}},
url = {http://dx.doi.org/10.1016/j.cpc.2013.03.020},
volume = 184,
year = 2012
}