Parallel computing. Efficient energy and power.

A project of research European, called repair and coordinated by the University Carlos III of Madrid (UC3M), studied how improve them applications computer of computing in parallel to increase its performance, efficiency energy and facilitate the programming and maintenance of the code source.
The computing heterogeneous in parallel combines several elements of processing of different features that share a unique system of memory. Normally multiple core processors (such as the ' multicore' of some smartphones or desktops) are used combined with graphics cards and other components to perform processing of large amounts of data.
These works of calculation found applications in diverse sectors, such as health (prediction of protein mesh), transport (railway systems monitoring), Robotics (stereoscopic vision and navigation) or industrial (analysis of defects in the manufacture of parts).
The objective of the project REPAIRED is put at the disposal of the users the benefits of performance of these computer systems, without the huge development effort involving this type of complex architectures and energy. And the key to achieve this lies, among other things, on "refactoring" source code, a program internal structure a technique used in software engineering to improve without changing its observable behavior. Something like change the distribution of pipes and pressure of a building so that the water comes out fast, clean and ecological way.
For repairs, whose denomination can interpret is as «Refactoring for Parallelism» in allusion to the use of technical of refactoring in it computing in parallel, them scientific try of improve three properties fundamental: the performance of them applications (helping to increase its speed of execution), it efficiency energy (reducing your consumption energy) and the ease of maintenance and modification of the code source.
"The tools and technologies that repair is to produce are intended to reduce development costs and time to market, which will result in more competitive products", says the Coordinator of the project, José Daniel García, of the research group of computer architecture, communications and systems (arcs) of UC3M.
Researchers hope to have prototypes of new tools to February 2015 and try them later in an assessment phase with industrial partners. "We hope to achieve important improvements, both in speed of execution and the energy efficiency of computers," advances the Professor José Daniel García. "And also can reduce the time that engineers need to develop a software product in a parallel environment, which notably clunker development costs and would promote the competitiveness of the industry of software development", highlights.
In a work of research published recently in the magazine New Generation Computing, the researchers have analyzed how affects the model of programming to the performance and the efficiency energy for architectures of processors of several nuclei. Starting from here precisely emerged the idea of delve in this line, because "is makes necessary help to them developers in the transition to different models of programming, as well as in the selection of the model of programming more suitable depending on them features concrete of its application", explains the Professor Jose Daniel Garcia.


Source: uc3m

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This site uses Akismet to reduce spam. Learn how your comment data is processed.