ON THE OBJECT MODELLING OF THE MASSIVELY PARALLEL ARCHITECTURE COMPUTERS

ABSTRACT In this paper, we present a modelling method to describe and emulate the massively parallel single Instruction Multiple Data (SIMD) structure machines. Among the modelled machines, we distinguish the linear, 2D meshes and pyramidal structures. All these computers are based physically on a multitude of fine grained processing elements (PE) arranged and coupled according to their associated topological pattern. Basing on the object modelling technique and on the XML description language, we develop a hard kernel of a parallel 2D virtual machine in which we translate all the physical properties of its different components. This kernel can be easily extended to the other mentioned topological structures. A parallel programming language and its compiler are also developed to edit, compile and run parallel programs. Furthermore, some illustrative examples will be given to show how the developed instruction sets can be manipulated to touch a large field of the parallel data processing applications.

[1]  Ramachandran Vaidyanathan,et al.  On the power of segmenting and fusing buses , 1993, [1993] Proceedings Seventh International Parallel Processing Symposium.

[2]  Omar Bouattane,et al.  θ(1) Time Algorithm for Structural Characterization of Multi-Leveled Images and its Applications on a Reconfigurable Mesh Computer , 2005, J. Intell. Robotic Syst..

[3]  Larry S. Davis,et al.  Efficient Parallel Processing of Image Contours , 1993, IEEE Trans. Pattern Anal. Mach. Intell..

[4]  Russ Miller,et al.  Mesh Computer Algorithms for Computational Geometry , 1989, IEEE Trans. Computers.

[5]  Vaidy S. Sunderam,et al.  Emulating Parallel Programming Environments in the Harness Metacomputing System , 2001, Parallel Process. Lett..

[6]  Omar Bouattane,et al.  A Fast Parallel Algorithm for Convex Hull Problem of Multi-Leveled Images , 2002, J. Intell. Robotic Syst..

[7]  O. Bouattane,et al.  Parallel algorithms for quadtree representation of points and curvilinear data , 1998 .

[8]  Vaidy S. Sunderam,et al.  PVM Emulation in the Harness Metacomputing Framework - Design and Performance Evaluation , 2002, 2nd IEEE/ACM International Symposium on Cluster Computing and the Grid (CCGRID'02).

[9]  Omar Bouattane,et al.  A Fast Algorithm for k-Nearest Neighbor Problem on a Reconfigurable Mesh Computer , 2001, J. Intell. Robotic Syst..

[10]  Mauro Migliardi,et al.  Parallel Component Descriptor Language: XML based deployment descriptors for Grid application components , 2007, PDPTA.

[11]  J. P. Gray,et al.  Configurable hardware: a new paradigm for computation , 1989 .

[12]  Massimo Maresca,et al.  Virtual Parallelism allows Relaxing the Synchronization Constraints of SIMD Computing Paradigm , 1998, HPCN Europe.

[13]  Omar Bouattane,et al.  θ(1) time quadtree algorithm and its application for image geometric properties on a mesh connected computer (MCC) , 1995, IEEE Transactions on Systems, Man, and Cybernetics.

[14]  Charles C. Weems,et al.  Image understanding architecture and applications , 1988 .

[15]  Massimo Maresca,et al.  Polymorphic-Torus Architecture for Computer Vision , 1989, IEEE Trans. Pattern Anal. Mach. Intell..