10396

SOCL: An OpenCL Implementation with Automatic Multi-Device Adaptation Support

Sylvain Henry, Denis Barthou, Alexandre Denis, Raymond Namyst, Marie-Christine Counilh
Sylvain Henry, Denis Barthou, Alexandre Denis, Raymond Namyst, Marie-Christine Counilh
hal-00853423, 22 August 2013

@techreport{henry:hal-00853423,

   hal_id={hal-00853423},

   url={http://hal.inria.fr/hal-00853423},

   title={SOCL: An OpenCL Implementation with Automatic Multi-Device Adaptation Support},

   author={Henry, Sylvain and Barthou, Denis and Denis, Alexandre and Namyst, Raymond and Counilh, Marie-Christine},

   keywords={OpenCL; Heterogeneous Architectures; Runtime Systems},

   language={Anglais},

   affiliation={Laboratoire Bordelais de Recherche en Informatique – LaBRI , RUNTIME – INRIA Bordeaux – Sud-Ouest},

   pages={18},

   type={Rapport de recherche},

   institution={INRIA},

   number={RR-8346},

   year={2013},

   month={Aug},

   pdf={http://hal.inria.fr/hal-00853423/PDF/RR-8346.pdf}

}

Download Download (PDF)   View View   Source Source   

1231

views

To fully tap into the potential of today’s heterogeneous machines, offloading parts of an application on accelerators is not sufficient. The real challenge is to build systems where the application would permanently spread across the entire machine, that is, where parallel tasks would be dynamically scheduled over the full set of available processing units. In this report we present SOCL, an OpenCL implementation that improves and simplifies the programming experience on heterogeneous architectures. SOCL enables applications to dynamically dispatch computation kernels over processing devices so as to maximize their utilization. OpenCL applications can incrementally make use of light extensions to automatically schedule kernels in a controlled manner on multi-device architectures. A preliminary automatic granularity adaptation extension is also provided. We demonstrate the relevance of our approach by experimenting with several OpenCL applications on a range of representative heterogeneous architectures. We show that performance portability is enhanced by using SOCL extensions.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: