Quantized load distribution for tree and bus-connected processors Gerassimos Barlas a, * , Bharadwaj Veeravalli b a Department of Computer Science, College of Engineering, American University of Sharjah, POB 26666, Sharjah, UAE b Department of Electrical and Computer Engineering, Open Source Software Laboratory, The National University of Singapore, 4 Engineering Drive 3, Singapore 117576, Singapore Received 15 July 2002; revised 15 April 2004; accepted 1 June 2004 Abstract Divisible load analysis is a valuable tool for generating solutions to data-partitioning and distribution/scheduling problems for data-parallel applications. This paper addresses an essen- tial step required for applying these solutions to real-life problems where computing loads are multiples of some fundamental problem-specific non-divisible load unit. The algorithms that are proposed to this end, are suitable for both single and multi-installment strategies. The worst-case performances of the algorithms are derived for two cases: single installment on a single-level tree and multiple installments on a bus network. Finally, an estimation on the ex- pected performance of the algorithms is obtained from rigorous simulation tests. The extensive analysis that accompanies these tests, depicts many aspects of the parallel computation asso- ciated with parallel machine architectures and the load distribution strategies (single- or multi- installment) used. Ó 2004 Elsevier B.V. All rights reserved. Keywords: Quantized load; Divisible load; Data partitioning; Processor trees; Bus-connected processors; Multi-installment 0167-8191/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.parco.2004.06.002 * Corresponding author. E-mail addresses: gbarlas@aus.ac.ae (G. Barlas), elebv@nus.edu.sg (B. Veeravalli). URL: http://opensource.nus.edu.sg. www.elsevier.com/locate/parco Parallel Computing 30 (2004) 841–865