Abstract

Today current era of scientific computing and computational theory involves high exhaustive data computation, shifted the trend of data processing from conventional processing towards parallel processing by incorporating multiple processing hardware. Parallel hardware design can employ array processors, pipelined system which can be further extended to scalar and super scalar pipelined systems. Other hardware designs proposed, is based upon multiprocessors or they may be designed as distributed parallel cluster systems. In this paper, multi-computers are the basic hardware for cluster design over the local area network covering analysis of matrix multiplication with strassen's algorithm. The estimated results are then compared with traditional matrix multiplication algorithm. Srassen's multiplication approach reduces one multiplication out of eight by computing arithmetic additions/subtractions for each 2×2 matrix. High performance can be achieved as the idea is extended over to multi-computer cluster for large sized matrices. This work covers analysis of Strassen's ability of divide and conquers[5] to run in parallel by decomposing matrix size over cluster machines covering data parallel aspects with SIMD based computational model [4], where each cluster machine performs its own recursive divide and conquer approach as defined by strassen's methodology[9][10] to obtain partitioned matrix multiplication. Finally, the detailed distributed experiment along with connectivity interface and implementation will be discussed.

References

• Ankit Arora, Amit Chhabra 2011 Cluster based Performance Evaluation of Run-length Image Compression Volume 33–No. 5, international Journal of Computer Applications, Foundation of Computer Science New York.
• Amit chhabra, Gurwinder Singh 2010 Cluster Based Parallel Computing framework for Performance evaluation of Parallel Applications, Vol. 2 April – 2, International Journal of Computer Theory and Engineering.
• Amit chhabra, Gurwinder Singh 2009 Simulated Performance Analysis of Multiprocessor Dynamic Space Sharing Scheduling Policy, Vol. 9 Feb – 2, International Journal of Computer Theory and Engineering
• Michael Sung, 2000 SIMD Parallel Processing, 6. 911: Architectures Anonymous
• Jonathan C. Hardwick 1997 "Practical Parallel Divide-and- Conquer Algorithms", CMU-CS-97-197
• Phyllis E. Crandall, Michael. j. Quinn 1993 Data Partitioning for Networked Parallel Processing, In Proceedings Data of Fifth Symposium on Parallel and Distributed Processing, Irving, TX.
• Visual Basic 6 Client/Server Programming Gold Book 1998, The Coriolis Group, ISBN: 1576102823.
• Bob Quinn Dec, 1995. Windows Socket Network Processing second edition Addison Wesley Professional, ISBN-10:0-201-633728.
• Ellis Horowitz, Sahni, 1978 Fundamentals of computer Algorithm Second edition, Computer science press.
• Thomas . H Coremen, Introduction to Algorithms, Second Edition Massachusetts institute of technology ISBN 0-262-03293-7.
• Omer Khan, MiesZko Lis, A Scalable Shared Memory Multicore Architecture. Massachusetts institute of technology Cambridge, MA, USA, june-2010, MIT-CSAIL-TR-2010-030.
• Steve J. Chapin, Syracuse University, Distributed and Multiprocessor Scheduling Volume 28 Issue 1, March 1996 ACM Digital Library New York, NY, USA.
• Kameswari Chebrolu, Socket Programming, Dept. of Electrical Engineering, IIT Kanpur.
• Rajinder Yadav, Client/Server Programming with TCP/IP Sockets, Sept 9, 2007.