This paper describes a task scheduling algorithm, based on a LogP -type model, for allocating arbitrary task graphs to fully connected networks of processors. This problem is known to be NP-complete even under the delay model (a special case under the LogP model). The strategy exploits the replication and clustering of tasks to minimise the ill e ects of communication overhead on the makespan. ...

This paper critique is written under the assumption that the reader is already familiar with the primary text [2] (henceforth referred to as “the paper”), and therefore a full overview of the LogP model will not be presented. The critique will mainly focus on shortcomings and weaknesses of the paper, leaving commentary on the positive contributions of the work for the end. The critique is organ...

A quantitative comparison of the BSP and LogP models of parallel computation is developed. We concentrate on a variant of LogP that disallows the so-called stalling behavior, although issues surrounding the stalling phenomenon are also explored. Very eecient cross simulations between the two models are derived, showing their substantial equivalence for algorithmic design guided by asymptotic an...

A quantitative comparison of the BSP and LogP models of parallel computation is developed. We concentrate on a variant of LogP that disallows the so-called stalling behavior, although issues surrounding the stalling phenomenon are also explored. Very eecient cross simulations between the two models are derived, showing their substantial equivalence for algorithmic design guided by asymptotic an...

Most theoretical work is based on the PRAM-model which has a block of shared memory and executes in a synchronous lock-step mode. Real hardware usually executes asynchronously and uses local memory and message passing. The recent LogP-model reeects these architectural properties. We show that for a practically important subclass of PRAM-programs it is possible to transform them into LogP-progra...

Currently, many parallel algorithms are deened for shared-memory architectures. The prefered machine model for designing these algorithms is the PRAM. However, this model does not take into account properties of existing architectures. Recently, Culler et al. deened the LogP machine model which better reeects the behaviour of massively parallel computers. We discuss an important class of progra...

It is shown that a threshold network of width w and depth d cannot be constructed from balancers of width p0; p1; : : : ; pm 1, if w does not divide P , where P is the least common multiple of p0; p1; : : : ; pm 1. This holds regardless of the size of the network, as long as it is nite, and it implies a lower bound of logP w on the depth of the network. More strongly, a lower bound of logpmax w...

The problem of load balancing on the hypercube is considered. A number of tokens are placed at the nodes and the goal is to redistribute them evenly throughout the network. Initially each of the p nodes stores up to m tokens. A simple algorithm is presented operating in time O(logp+m log logp) on average.

In this paper, the LogP model is used to analyze four parallel sorting algorithms (bitonic, column, radix, and sample sort). LogP characterizes the performance of modern parallel machines with a small set of parameters: the communication latency (L), overhead (o), bandwidth (g), and the number of processors (P ). We develop implementations of these algorithms in Split-C, a parallel extension to...