Task tree executor for Erlang enabled multicores

Abstract

Task trees are important class of parallel programs that are found in a broad range of applications, from critical infrastructures such as electricity, oil, and gas distribution networks, to modern large-scale computer games. Erlang is a modern parallel programming language offering efficient fine-grain parallelism, based on its ultra light processes, and direct executable portability on various multicores. The goal of the work presented in this paper was to develop the novel Task Tree Executor (TTE) for Erlang enabled multicores. After introducing TTE task trees we present the novel Erlang based TTE. Then, we construct DAG model for TTE binary trees and derive formulas for parallelism and lower bound of speedup. Finally, we validate the proposed theory by experiments. © 2013 American Scientific Publishers All rights reserved.