M. Adler, J. Byers, R. Karp. Parallel Sorting with Limited Bandwidth Proc. 7th ACM Symp. on Parallel Algorithms and Architectures: pp. 129 - 136, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of m bits. We provide new algorithms for the problems of sorting and permutation routing. For the concurrent read PRAM(m) as p grows with m held constant, our sorting algorithm outperforms any previous algorithm by Omega Gamma409 c p) for any constant c. The combination of a lower bound from [ABK95] for sorting in the exclusive read PRAM(m) model and this algorithm implies that the concurrent read PRAM(m) is strictly more powerful than the exclusive read PRAM(m) 1.1 An I O Complexity Application We first describe an application of our technique in a simplified version of Aggarwal and ....

.... technique in this paper is reducing the bandwidth requirements of algorithms in the PRAM(m) a limited bandwidth parallel model introduced by [MNV94] Low communication throughput has been shown to be a primary bottleneck in parallel computation by several results, including [VW85] MNV94] [ABK95], and [Goo96] Thus, in many situations, using the available bandwidth as efficiently as possible is crucial. In the PRAM(m) p processors communicate only through a small, globally shared memory consisting of m bits. The input, assumed to be of size n, is stored in a globally readable ROM. ....

[Article contains additional citation context not shown here]

M. Adler, J. Byers, R. Karp. Parallel Sorting with Limited Bandwidth Proc. 7th ACM Symp. on Parallel Algorithms and Architectures: pp. 129 - 136, 1995.

....n log n m bits of the PRAM memory, we can run Cole s algorithm in time O( n log 2 n m ) on the ER PRAM(m) Thus, the ratio between upper and lower bounds for sorting on the ER PRAM(m) was Theta( q p m log 2 n) prior to this work. Subsequent to a preliminary version of this paper in [ABK95] Adler [Adl96] provides an algorithm for sorting in the CR PRAM(m) that is considerably faster than the lower bound for the ER PRAM(m) presented in this paper. Thus, that result together with the lower bound presented here imply that the CR PRAM(m) is strictly more powerful than the ER PRAM(m) ....

....consists of exactly one key. Their algorithms run in time O( n log n P gp ffl gn P L) with high probability, for any positive constant ffl 1, and for P n 1 Gammaffi , where ffi is a small constant depending on ffl. After the preliminary version of this paper appeared in [ABK95] work on this problem by Goodrich [Goo96] has tightened the bounds for sorting on the BSP, giving deterministic algorithms which run in time O( n log n P (L gn P ) log n= log(n=P ) for all values of P , coupled with a matching lower bound. Other recent work by Gerbessiotis and ....

M. Adler, J. Byers, R. Karp. Parallel Sorting with Limited Bandwidth Proc. 7th ACM Symp. on Parallel Algorithms and Architectures: pp. 129 - 136, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC