HPC² - A fully-portable, algebra-based framework for heterogeneous computing. Application to CFD

The variety of computing architectures competing in the exascale race makes the portability of codes of major importance. In this work, the HPC2 code is presented as a fully-portable, algebra-based framework suitable for heterogeneous computing. In its application to CFD, the algorithm of the time-i...

Descripción completa

Detalles Bibliográficos
Autores: Álvarez Farré, Xavier|||0000-0002-1684-7658, Gorobets, Andrei, Trias Miquel, Francesc Xavier|||0000-0002-5966-0703, Borrell Pol, Ricard, Oyarzun Altamirano, Guillermo|||0000-0001-9524-3782
Tipo de recurso: artículo
Fecha de publicación:2018
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/116810
Acceso en línea:https://hdl.handle.net/2117/116810
https://dx.doi.org/10.1016/j.compfluid.2018.01.034
Access Level:acceso abierto
Palabra clave:Computational fluid dynamics
Turbulence--Computer simulation
Heterogeneous computing
MPI+OpenMP+OpenCL
Hybrid CPU+GPU systems
CFD
Symmetry-preserving discretization
Dinàmica de fluids computacional
Turbulència -- Simulació numèrica
Àrees temàtiques de la UPC::Enginyeria mecànica::Mecànica de fluids
Descripción
Sumario:The variety of computing architectures competing in the exascale race makes the portability of codes of major importance. In this work, the HPC2 code is presented as a fully-portable, algebra-based framework suitable for heterogeneous computing. In its application to CFD, the algorithm of the time-integration phase relies on a reduced set of only three algebraic operations: the sparse matrix-vector product, the linear combination of vectors and the dot product. This algebraic approach combined with a multilevel MPI+OpenMP+OpenCL parallelization naturally provides portability. The performance has been studied on different architectures including multicore CPUs, Intel Xeon Phi accelerators and GPUs of AMD and NVIDIA. The multi-GPU scalability is demonstrated up to 256 devices. The heterogeneous execution is tested on a CPU+GPU hybrid cluster. Finally, results of the direct numerical simulation of a turbulent flow in a 3D air-filled differentially heated cavity are presented to show the capabilities of the HPC2 dealing with large-scale CFD simulations.