This paper introduces a new programming control structure called "xfor" as an extension of the classical "for" construct in C. It is designed to help one programmer to improve data locality on multi-core architectures by allowing him to express the schedule of instructions in an abstract way. This schedule is defined geometrically by mapping the iteration domains relatively to each other onto a unique referential by using specific parameters called grain and offset. A semantic framework is presented which associates a precise meaning with this syntactic construct and serves as a base for applying reliable xfor code transformations and programming strategies. These issues are illustrated with the Red-Black algorithm. Performance measurements...
There is an increasing need for a framework that supports research on portable high-performance para...
263 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1997.The increasing amount of inst...
International audienceImproving execution time and energy efficiency is needed for many applications...
This paper introduces a new programming control structure called "xfor" as an extension of the class...
International audienceWe propose a new loop structure named {\em xfor}, offering programmers explici...
We propose a new programming structure named "XFOR" or "Multifor", dedicated to data-reuse aware pro...
We propose a new programming structure named XFOR (Multifor), dedicated to data-reuse aware programm...
Nous proposons une nouvelle structure de programmation appelée XFOR (Multifor), dédiée à la programm...
International audienceData locality optimization is a well-known goal when handling programs that mu...
International audienceDevelopment of an HPC simulation code may take years of a domain scientists' w...
In this technical report, we present information on the XMTC compiler and language. We start by pre...
International audienceX10 is a promising recent parallel language designed specifically to address t...
This work focuses on compiler and run-time techniques for improving the productivity and the perform...
The ability to represent, manipulate and optimize data placement and movement between processors in ...
International audienceWe propose a new programming control structure called ''multifor'', allowing t...
There is an increasing need for a framework that supports research on portable high-performance para...
263 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1997.The increasing amount of inst...
International audienceImproving execution time and energy efficiency is needed for many applications...
This paper introduces a new programming control structure called "xfor" as an extension of the class...
International audienceWe propose a new loop structure named {\em xfor}, offering programmers explici...
We propose a new programming structure named "XFOR" or "Multifor", dedicated to data-reuse aware pro...
We propose a new programming structure named XFOR (Multifor), dedicated to data-reuse aware programm...
Nous proposons une nouvelle structure de programmation appelée XFOR (Multifor), dédiée à la programm...
International audienceData locality optimization is a well-known goal when handling programs that mu...
International audienceDevelopment of an HPC simulation code may take years of a domain scientists' w...
In this technical report, we present information on the XMTC compiler and language. We start by pre...
International audienceX10 is a promising recent parallel language designed specifically to address t...
This work focuses on compiler and run-time techniques for improving the productivity and the perform...
The ability to represent, manipulate and optimize data placement and movement between processors in ...
International audienceWe propose a new programming control structure called ''multifor'', allowing t...
There is an increasing need for a framework that supports research on portable high-performance para...
263 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1997.The increasing amount of inst...
International audienceImproving execution time and energy efficiency is needed for many applications...