Desenvolvimento de uma Versao Paralela do PhyML 3.0 Usando Memoria Compartilhada (Development of a Parallel Version of PhyML 3.0 Using Shared Memory)

Julio Oliveira da Silva (julioios@hotmail.com)0, Esbel Tomas Valero Orellana (evalero@uesc.br)2, Martha Ximena Torres Delgado (mxtd2000@gmail.com)2


1
2Universidade Estadual de Santa Cruz

This paper appears in: Revista IEEE América Latina

Publication Date: May 2017
Volume: 15,   Issue: 5 
ISSN: 1548-0992


Abstract:
This work presents the main steps towards a parallel version of the open source phylogenetic tree reconstruction software PhyML 3.0, which uses the method of maximum likelihood in phylogenetic reconstruction process. The PhyML was parallelized by a shared memory approach (using OpenMP) and at the same time using the version already consolidated in PhyML 3.0, the version passing messages with distributed memory (using MPI), was also constructed a version hybrid (MPI and OpenMP) of PhyML. The tests for the efficiency and speedup were carried out in a cluster of two ways, the first only by analyzing the performance version with OpenMP (DNA and proteins) and the second making a comparative analysis of the performance of three parallel versions (OpenMP, MPI and Hybrid) using DNA data only. The results obtained show speedup of approximately six to 8 cores on 1 server, which allows the use of this code problems that require a lot of time in the serial version, front the need or urgent of the fast processing of data.

Index Terms:
Bioinformatics, High-performance computing, Phylogenetic Tree Reconstruction, OpenMP, MPI   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (652)]