Pascal Bouvry

Efficient Solutions for Mapping Parallel Programs

This paper describes a mapping toolbox, whose aim is to optimize the execution time of parallel programs described as task graphs on distributed memory parallel systems. The toolbox includes several classical mapping algorithms. It was assessed by computing the mapping of randomly generated task graphs and by mapping and executing on a parallel system synthetic programs representing some classical numerical algorithms. A large number of experiments were used to validate the cost functions used in the toolbox and to compare the algorithms.

ANDES: evaluating mapping strategies with synthetic programs

Abstract This paper presents the ANDES performance evaluation tool. ANDES is based on the synthetic execution of parallel programs and it is used for the evaluation of mapping strategies. The Meganode, a distributed memory parallel computer, is considered as our target architecture. ANDES takes into account a benchmark of quantitative models of parallel algorithms and a set of mapping strategies (greedy and iterative algorithms are used). We show how this tool allows an extensive comparison of mapping strategies by using the benchmark, the mapping strategies and different cost functions.

Scheduling complete intrees on two uniform processors with communication delays

We present an optimal algorithm for scheduling a complete k-ary tree on two uniform processors of different speeds in order to minimize schedule length. We consider the basic case of unit standard execution times and unit communication times.