A Combined MPI-CUDA Parallel Solution of Linear and Nonlinear Poisson-Boltzmann Equation

• Published in: BioMed research international Vol. 2014; no. 2014; pp. 1 - 12
• Main Authors: Colmenares, José, Galizia, Antonella, Ortiz, Jesús, Clematis, Andrea, Rocchia, Walter
• Format: Journal Article
• Language: English
• Published: Cairo, Egypt Hindawi Puplishing Corporation 01.01.2014; Hindawi Publishing Corporation; John Wiley & Sons, Inc
• Subjects: Algorithms; Applications software; Biological; Biomedical research; Biomolecules; Computation; Computational Biology; Computer programs; Computer Simulation; Electrolytes; Electrostatics; Field programmable gate arrays; Mathematical analysis; Mathematical models; Models, Molecular; Obstacles; Protein Conformation; Proteins - chemistry; Software; Solvents; Static Electricity; Studies; Italy
• ISSN: 2314-6133; 2314-6141; 2314-6141
• DOI: 10.1155/2014/560987

The Poisson-Boltzmann equation models the electrostatic potential generated by fixed charges on a polarizable solute immersed in an ionic solution. This approach is often used in computational structural biology to estimate the electrostatic energetic component of the assembly of molecular biological systems. In the last decades, the amount of data concerning proteins and other biological macromolecules has remarkably increased. To fruitfully exploit these data, a huge computational power is needed as well as software tools capable of exploiting it. It is therefore necessary to move towards high performance computing and to develop proper parallel implementations of already existing and of novel algorithms. Nowadays, workstations can provide an amazing computational power: up to 10 TFLOPS on a single machine equipped with multiple CPUs and accelerators such as Intel Xeon Phi or GPU devices. The actual obstacle to the full exploitation of modern heterogeneous resources is efficient parallel coding and porting of software on such architectures. In this paper, we propose the implementation of a full Poisson-Boltzmann solver based on a finite-difference scheme using different and combined parallel schemes and in particular a mixed MPI-CUDA implementation. Results show great speedups when using the two schemes, achieving an 18.9x speedup using three GPUs.