TY - JOUR
T1 - Optimization of finite-differencing kernels for numerical relativity applications
AU - Alfieri, Roberto
AU - Bernuzzi, Sebastiano
AU - Perego, Albino
AU - Radice, David
N1 - Publisher Copyright:
© 2018 by the authors. Licensee MDPI, Basel, Switzerland.
PY - 2018/6
Y1 - 2018/6
N2 - A simple optimization strategy for the computation of 3D finite-differencing kernels on many-cores architectures is proposed. The 3D finite-differencing computation is split direction-by-direction and exploits two level of parallelism: in-core vectorization and multi-threads shared-memory parallelization. The main application of this method is to accelerate the high-order stencil computations in numerical relativity codes. Our proposed method provides substantial speedup in computations involving tensor contractions and 3D stencil calculations on different processor microarchitectures, including Intel Knight Landing.
AB - A simple optimization strategy for the computation of 3D finite-differencing kernels on many-cores architectures is proposed. The 3D finite-differencing computation is split direction-by-direction and exploits two level of parallelism: in-core vectorization and multi-threads shared-memory parallelization. The main application of this method is to accelerate the high-order stencil computations in numerical relativity codes. Our proposed method provides substantial speedup in computations involving tensor contractions and 3D stencil calculations on different processor microarchitectures, including Intel Knight Landing.
UR - http://www.scopus.com/inward/record.url?scp=85048307655&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85048307655&partnerID=8YFLogxK
U2 - 10.3390/jlpea8020015
DO - 10.3390/jlpea8020015
M3 - Article
AN - SCOPUS:85048307655
SN - 2079-9268
VL - 8
JO - Journal of Low Power Electronics and Applications
JF - Journal of Low Power Electronics and Applications
IS - 2
M1 - 15
ER -