Tetsu Narumi

42 TFlops hierarchical N -body simulations on GPUs with applications in both astrophysics and turbulence

As an entry for the 2009 Gordon Bell price/performance prize, we present the results of two different hierarchical N-body simulations on a cluster of 256 graphics processing units (GPUs). Unlike many previous N-body simulations on GPUs that scale as O(N2), the present method calculates the O(N log N) treecode and O(N) fast multipole method (FMM) on the GPUs with unprecedented efficiency. We demonstrate the performance of our method by choosing one standard application -a gravitational N-body simulation- and one non-standard application -simulation of turbulence using vortex particles. The gravitational simulation using the treecode with 1,608,044,129 particles showed a sustained performance of 42.15 TFlops. The vortex particle simulation of homogeneous isotropic turbulence using the periodic FMM with 16,777,216 particles showed a sustained performance of 20.2 TFlops. The overall cost of the hardware was 228,912 dollars. The maximum corrected performance is 28.1TFlops for the gravitational simulation, which results in a cost performance of 124 MFlops/

High-performance drug discovery: computational screening by combining docking and molecular dynamics simulations.

. This correction is performed by counting the Flops based on the most efficient CPU algorithm. Any extra Flops that arise from the GPU implementation and parameter differences are not included in the 124 MFlops/

Molecular Dynamics Machine: Special-Purpose Computer for Molecular Dynamics Simulations

.

Fast multipole methods on a cluster of GPUs for the meshless simulation of turbulence

Virtual compound screening using molecular docking is widely used in the discovery of new lead compounds for drug design. However, this method is not completely reliable and therefore unsatisfactory. In this study, we used massive molecular dynamics simulations of protein-ligand conformations obtained by molecular docking in order to improve the enrichment performance of molecular docking. Our screening approach employed the molecular mechanics/Poisson-Boltzmann and surface area method to estimate the binding free energies. For the top-ranking 1,000 compounds obtained by docking to a target protein, approximately 6,000 molecular dynamics simulations were performed using multiple docking poses in about a week. As a result, the enrichment performance of the top 100 compounds by our approach was improved by 1.6–4.0 times that of the enrichment performance of molecular dockings. This result indicates that the application of molecular dynamics simulations to virtual screening for lead discovery is both effective and practical. However, further optimization of the computational protocols is required for screening various target proteins.

DS-CUDA: A Middleware to Use Many GPUs in the Cloud Environment

Abstract We are now developing Molecular Dynamics Machine (MDM), a special-purpose computer for classical molecular dynamics simulations. It accelerates the calculation of non-bonding force, Coulomb and van der Waals forces, because the calculation cost for Coulomb force dominates the total calculation time when we treat a large system of charged particles without truncating Coulomb force. When we use Ewald method, the Coulomb force can be calculated by dividing it into real-space and wavenumber-space parts. MDM is composed of MDGRAPE-2, WINE-2, and a host computer. MDGRAPE-2 calculates van der Waals force and real-space part of Coulomb force. WINE-2 calculates wavenumber-space part of Coulomb force. The host computer calculates bonding-force and updates positions and velocities of atoms. The target performance of MDM is 100 Tflops and will sustain about 30 Tflops in realistic applications. It can calculate 3.2 × 106 time-steps of MD simulation with a million atoms in a week. Total system will be complete...

Cutoff radius effect of the isotropic periodic sum and Wolf method in liquid-vapor interfaces of water.

Recent advances in the parallelizability of fast N-body algorithms, and the programmability of graphics processing units (GPUs) have opened a new path for particle based simulations. For the simulation of turbulence, vortex methods can now be considered as an interesting alternative to finite difference and spectral methods. The present study focuses on the efficient implementation of the fast multipole method and pseudo-particle method on a cluster of NVIDIA GeForce 8800 GT GPUs, and applies this to a vortex method calculation of homogeneous isotropic turbulence. The results of the present vortex method agree quantitatively with that of the reference calculation using a spectral method. We achieved a maximum speed of 7.48 TFlops using 64 GPUs, and the cost performance was near

Molecular dynamics simulations of vapor/liquid coexistence using the nonpolarizable water models

9.4/GFlops. The calculation of the present vortex method on 64 GPUs took 4120 s, while the spectral method on 32 CPUs took 4910 s.

Novel Mechanism of Interaction of p85 Subunit of Phosphatidylinositol 3-Kinase and ErbB3 Receptor-derived Phosphotyrosyl Peptides

GPGPU (General-purpose computing on graphics processing units) has several difficulties when used in cloud environment, such as narrow bandwidth, higher cost, and lower security, compared with computation using only CPUs. Most high performance computing applications require huge communication between nodes, and do not fit a cloud environment, since network topology and its bandwidth are not fixed and they affect the performance of the application program. However, there are some applications for which little communication is needed, such as molecular dynamics (MD) simulation with the replica exchange method (REM). For such applications, we propose DS-CUDA (Distributed-shared compute unified device architecture), a middleware to use many GPUs in a cloud environment with lower cost and higher security. It virtualizes GPUs in a cloud such that they appear to be locally installed GPUs in a client machine. Its redundant mechanism ensures reliable calculation with consumer GPUs, which reduce the cost greatly. It also enhances the security level since no data except command and data for GPUs are stored in the cloud side. REM-MD simulation with 64 GPUs showed 58 and 36 times more speed than a locally-installed GPU via InfiniBand and the Internet, respectively.

Thermodynamic properties of methane/water interface predicted by molecular dynamics simulations

As a more economical but similarly accurate computation method than the Ewald sum, the isotropic periodic sum (IPS) method for nonpolar molecules (IPSn) and polar molecules (IPSp), along with the Wolf method are of interest, but the cutoff radius dependence is an important issue. To evaluate the cutoff radius effect of the three methods, a water-vapor interfacial system has been studied by molecular dynamics. The Wolf method can produce adequate results for surface tension compared to that of the Ewald sum (within 2.9%) at a long enough cutoff radius, r(c). However, the estimation of the electrostatic potential profile and dipole orientational function is poor. The Wolf method cannot estimate electrostatic configuration at r(c) ≤ L(z)∕2 (L(z) is the longest lattice of the system). We have found that the convergence of the surface tension and the electrostatic configuration of the IPSn method is faster than that of the IPSp method. Moreover, the IPSn method is most accurate among the three methods for the same cutoff radius. Furthermore, the behavior of the surface tension against the cutoff radius shows a greater difference for the IPSn and IPSp method. The surface tension of the IPSp method fluctuates and presents a similar result to that of the Ewald sum, but the surface tension for the IPSn method greatly deviates near r(c) = L(z)∕3. The cause of this deviation is the difference between the interfacial configuration of the water surface and the cutoff treatment of the IPS method. The deviation becomes insignificant far from r(c) = L(z)∕3. In spite of this shortcoming, the IPSn method gives the most accurate result in estimating the surface tension at r(c) = L(z)∕2. From all the results in this work, the IPSn and IPSp method have been found to be more accurate than the Wolf method. In conclusion, the surface tension and structure of water-vapor interface can be calculated by the IPSn method when r(c) is greater than or equal to the longest lattice of the system. The IPSp method and the Wolf method require a longer cutoff radius than the longest lattice of the system to estimate interfacial properties.

Cutoff radius effect of the isotropic periodic sum method in homogeneous system. II. Water

The surface tension, vapor-liquid equilibrium densities, and equilibrium pressure for common water models were calculated using molecular dynamics simulations over temperatures ranging from the melting to the critical points. The TIP4P/2005 and TIP4P-i models produced better values for the surface tension than the other water models. We also examined the correlation of the data to scaling temperatures based on the critical and melting temperatures. The reduced temperature (T/T(c)) gives consistent equilibrium densities and pressure, and the shifted temperature T + (T(c, exp) - T(c, sim)) gives consistent surface tension among all models considered in this study. The modified fixed charge model which has the same Lennard-Jones parameters as the TIP4P-FQ model but uses an adjustable molecular dipole moment is also simulated to find the differences in the vapor-liquid coexistence properties between fixed and fluctuating charge models. The TIP4P-FQ model (2.72 Debye) gives the best estimate of the experimental surface tension. The equilibrium vapor density and pressure are unaffected by changes in the dipole moment as well as the surface tension and liquid density.