Title :
The data parallelism under java paralleltask: A simple and efficient approach
Author :
Djema, Lounes ; Rassoul, Idir ; Boumghar, Fatima Oulebsir
Author_Institution :
Comput. Sci. Dept., Univ. UMMTO, Tizi-ouzou, Algeria
Abstract :
With the advent of multicore general public systems, a new perspective of parallel processing is available to application developers. Adapting old sequential applications to performance of parallel systems proves imperative. The Task Parallel Java compiler (ParaTask) is a solution for the development of object-oriented parallel applications, based on the concept of different types of tasks. It is easily exploitable for the change (refactoring) of existing sequential applications by using simple keywords. Parallelism is introduced either in the form of parallel data and parallel task, or as a mixed parallel / Data tasks. It comes to performed in this work, the parallelization of the sequential algorithm of Canny contour detection on images of different sizes.
Keywords :
Java; image representation; multiprocessing systems; object-oriented methods; parallel algorithms; program compilers; Canny contour image detection; ParaTask; data parallelism; mixed parallel-data tasks; multicore general public systems; object-oriented parallel applications; parallel processing; sequential algorithm parallelization; sequential applications; task parallel Java compiler; Image edge detection; Instruction sets; Java; Multicore processing; Multitasking; Parallel processing; ParaTask; cyclic data; data blocks; multicore; parallel; refactoring;
Conference_Titel :
Computer Applications and Information Systems (WCCAIS), 2014 World Congress on
Conference_Location :
Hammamet
Print_ISBN :
978-1-4799-3350-1
DOI :
10.1109/WCCAIS.2014.6916617