![]() |
|||
|
Thursday November 12, 2009. 11:00 am
Room 302 HRBB
Toward automatic partitioning of C applications for exploiting hardware accelerators such as GPUs
Francois Bodin.
CAPS entreprise
Abstract
Automatically exploiting hardware accelerators such as GPUs attached to a main CPU unit (i.e. the host) is a very complex task that requires the partitioning of an application in order to identify computation intensive hotspots that can benefit from the accelerator. Because hardware accelerators usually have their own address space, using them requires the identification of the data to move into accelerator memory, as well as taking into account data transfers overheads between the host and the accelerators. Usually, accelerators with very high performance are the results of highly parallel and pipelined architectures. Optimizing for such architecture relies on the usual program transformations (e.g. loop unrolling, etc.) but the optimization strategy differs significantly from those corresponding to a superscalar microprocessor.
In this presentation we give an overview of a set of speculative techniques for automatically partitioning C programs.
Biography
Francois Bodin cofounded CAPS entreprise in 2002 while he was a Professor at University of Rennes I and since January 2008 he joined the company as CTO. His contribution includes new approaches for exploiting high performance processors in scientific computing and in embedded applications. Prior to joining CAPS, Francois Bodin held various research positions at University of Rennes I and at the INRIA research lab.
Parasol Home | Research | People | General info | Seminars | Resources Parasol Lab, 301 Harvey R. Bright Bldg, 3112 TAMU, College Station, TX 77843-3112 Contact Webmaster Phone 979.458.0722 Fax 979.458.0718
Department of Computer Science and Engineering | Dwight Look College of Engineering | Texas A&M University Privacy statement: Computer Science and Engineering Engineering TAMU |