Performance Analysis of FEM Algorithmson GPU and Many-Core Architectures

Handle URI:
http://hdl.handle.net/10754/581405
Title:
Performance Analysis of FEM Algorithmson GPU and Many-Core Architectures
Authors:
Khurram, Rooh; Kortas, Samuel
Abstract:
The roadmaps of the leading supercomputer manufacturers are based on hybrid systems, which consist of a mix of conventional processors and accelerators. This trend is mainly due to the fact that the power consumption cost of the future cpu-only Exascale systems will be unsustainable, thus accelerators such as graphic processing units (GPUs) and many-integrated-core (MIC) will likely be the integral part of the TOP500 (http://www.top500.org/) supercomputers, beyond 2020. The emerging supercomputer architecture will bring new challenges for the code developers. Continuum mechanics codes will particularly be affected, because the traditional synchronous implicit solvers will probably not scale on hybrid Exascale machines. In the previous study[1], we reported on the performance of a conjugate gradient based mesh motion algorithm[2]on Sandy Bridge, Xeon Phi, and K20c. In the present study we report on the comparative study of finite element codes, using PETSC and AmgX solvers on CPU and GPUs, respectively [3,4]. We believe this study will be a good starting point for FEM code developers, who are contemplating a CPU to accelerator transition.
KAUST Department:
KAUST Supercomputing Laboratory (KSL)
Conference/Event name:
Pan American Congresses on Computational Mechanics (PANACM) 2015
Issue Date:
27-Apr-2015
Type:
Presentation
Additional Links:
http://congress.cimne.com/panacm2015/admin/files/fileabstract/a307.pdf
Appears in Collections:
KAUST Supercomputing Laboratory (KSL); Presentations

Full metadata record

DC FieldValue Language
dc.contributor.authorKhurram, Roohen
dc.contributor.authorKortas, Samuelen
dc.date.accessioned2015-10-29T14:04:08Zen
dc.date.available2015-10-29T14:04:08Zen
dc.date.issued2015-04-27en
dc.identifier.urihttp://hdl.handle.net/10754/581405en
dc.description.abstractThe roadmaps of the leading supercomputer manufacturers are based on hybrid systems, which consist of a mix of conventional processors and accelerators. This trend is mainly due to the fact that the power consumption cost of the future cpu-only Exascale systems will be unsustainable, thus accelerators such as graphic processing units (GPUs) and many-integrated-core (MIC) will likely be the integral part of the TOP500 (http://www.top500.org/) supercomputers, beyond 2020. The emerging supercomputer architecture will bring new challenges for the code developers. Continuum mechanics codes will particularly be affected, because the traditional synchronous implicit solvers will probably not scale on hybrid Exascale machines. In the previous study[1], we reported on the performance of a conjugate gradient based mesh motion algorithm[2]on Sandy Bridge, Xeon Phi, and K20c. In the present study we report on the comparative study of finite element codes, using PETSC and AmgX solvers on CPU and GPUs, respectively [3,4]. We believe this study will be a good starting point for FEM code developers, who are contemplating a CPU to accelerator transition.en
dc.relation.urlhttp://congress.cimne.com/panacm2015/admin/files/fileabstract/a307.pdfen
dc.titlePerformance Analysis of FEM Algorithmson GPU and Many-Core Architecturesen
dc.typePresentationen
dc.contributor.departmentKAUST Supercomputing Laboratory (KSL)en
dc.conference.date27-29 April, 2015en
dc.conference.namePan American Congresses on Computational Mechanics (PANACM) 2015en
dc.conference.locationBuenos Airesen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.