The Community for Technology Leaders
RSS Icon
Subscribe
Issue No.08 - August (1995 vol.6)
pp: 896-900
ABSTRACT
<p><it>Abstract</it>—Consider a message-passing system of <it>n</it> processors, in which each processor holds one piece of data initially. The goal is to compute an associative and commutative reduction function on the <it>n</it> pieces of data and to make the result known to all the <it>n</it> processors. This operation is frequently used in many message-passing systems and is typically referred to as <it>global combine, census computation,</it> or <it>gossiping</it>. This paper explores the problem of global combine in the <it>multiport postal model</it>. This model is characterized by three parameters: <it>n</it>—the number of processors, <it>k</it>—the number of ports per processor, and λ—the communication latency. In this model, in every round <it>r</it>, each processor can send <it>k</it> distinct messages to <it>k</it> other processors, and it can receive <it>k</it> messages that were sent from <it>k</it> other processors λ− 1 rounds earlier. This paper provides an optimal algorithm for the global combine problem that requires the least number of communication rounds and minimizes the time spent by any processor in sending and receiving messages.</p>
INDEX TERMS
Census computation, distributed systems, global combine, gossiping, message-passing systems, multiple ports, parallel computers, postal model.
CITATION
Amotz Bar-Noy, Jehoshua Bruck, Ching-Tien Ho, Shlomo Kipnis, Baruch Schieber, "Computing Global Combine Operations in the Multiport Postal Model", IEEE Transactions on Parallel & Distributed Systems, vol.6, no. 8, pp. 896-900, August 1995, doi:10.1109/71.406965
25 ms
(Ver 2.0)

Marketing Automation Platform Marketing Automation Tool