Impact of multicores on large-scale molecular dynamics simulations

Alam, Sadaf R.; Agarwal, Pratul K.; Hampton, Scott; Ong, Hong; Vetter, Jeffrey S.

doi:10.1109/ipdps.2008.4536181

Cited by 9 publications

(6 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is important to make the distinction between individual multi-core nodes and SMP systems. Multi-core processors often involve some level of cache sharing, which can have performance implications for parallel codes [1].…”

Section: Introductionmentioning

confidence: 99%

Performance analysis of a hybrid MPI/OpenMP application on multi-core clusters

Chorley

Walker

2010

Journal of Computational Science

View full text Add to dashboard Cite

a b s t r a c tThe mixing of shared memory and message passing programming models within a single application has often been suggested as a method for improving scientific application performance on clusters of shared memory or multi-core systems. DL POLY, a large scale molecular dynamics application programmed using message passing programming has been modified to add a layer of shared memory threading and the performance analysed on two multi-core clusters. At lower processor numbers, the extra overheads from shared memory threading in the hybrid code outweigh performance benefits gained over the pure MPI code. On larger core counts the hybrid model performs better than pure MPI, with reduced communication time decreasing the overall runtime.

show abstract

Section: Introductionmentioning

confidence: 99%

Performance analysis of a hybrid MPI/OpenMP application on multi-core clusters

Chorley

Walker

2010

Journal of Computational Science

View full text Add to dashboard Cite

show abstract

“…To quantitatively analyze the amount of coverage obtained by our method when compared to a single MD simulation, we present the following calculation: A 1 ns of MD simulation as performed in our previous work18 and in this work requires ∼6 h using 64 CPUs on a Cray XT3 machine. A single 20 ns MD simulation would therefore require a total of 120 h. Using the method outlined in this article, we can utilize 640 processors to invoke 10 simultaneous 2 ns MD simulations in 12 h. Using all 640 processors in a single MD simulation trying to get the same effect will not result in linear speedup due to the nature of the MPI protocol used by standard MD packages 20, 34. This way we achieve a practical linear speedup as a function of the number of processors, at the cost of losing time‐related information which can be achieved by a single, long MD simulation.…”

Section: Resultsmentioning

confidence: 99%

Multi‐scale characterization of the energy landscape of proteins with application to the C3D/Efb‐C complex

et al. 2009

View full text Add to dashboard Cite

We present a novel multi-level methodology to explore and characterize the low energy landscape and the thermodynamics of proteins. Traditional conformational search methods typically explore only a small portion of the conformational space of proteins and are hard to apply to large proteins due to the large amount of calculations required. In our multi-scale approach, we first provide an initial characterization of the equilibrium state ensemble of a protein using an efficient computational conformational sampling method. We then enrich the obtained ensemble by performing short Molecular Dynamics (MD) simulations on selected conformations from the ensembles as starting points. To facilitate the analysis of the results we project the resulting conformations on a low-dimensional landscape to efficiently focus on important interactions and examine low energy regions. This methodology provides a more extensive sampling of the low energy landscape than an MD simulation starting from a single crystal structure as it explores multiple trajectories of the protein. This enables us to obtain a broader view of the dynamics of proteins and it can help in understanding complex binding, improving docking results and more. In this work we apply the methodology to provide an extensive characterization of the bound complexes of the C3d fragment of human Complement component C3 and one of its powerful bacterial inhibitors, the inhibitory domain of Staphylococcus aureus extra-cellular fibrinogenbinding domain (Efb-C) and two of its mutants. We characterize several important interactions along the binding interface and define low free energy regions in the three complexes.

show abstract

“…There have been extensive interest in MD parallelization of academic and industry software packages regarding redesigned algorithms [13][14][15][16] for multicore CPUs [17][18][19][20] and computer clusters. [21][22][23][24] Publications approach the topic in various ways, most of them implementing parallel force calculations, but solvent effect computation and integration were also investigated.…”

Section: Introductionmentioning

confidence: 99%

Classical molecular dynamics on graphics processing unit architectures

Jász¹,

Rák²,

Ladjánszki³

et al. 2019

WIREs Comput Mol Sci

View full text Add to dashboard Cite

Molecular dynamics (MD) has experienced a significant growth in the recent decades. Simulating systems consisting of hundreds of thousands of atoms is a routine task of computational chemistry researchers nowadays. Thanks to the straightforwardly parallelizable structure of the algorithms, the most promising method to speed‐up MD calculations is exploiting the large‐scale processing power offered by the parallel hardware architecture of graphics processing units or GPUs. Programming GPUs is becoming easier with general‐purpose GPU computing frameworks and higher levels of abstraction. In the recent years, implementing MD simulations on graphics processors has gained a large interest, with multiple popular software packages including some form of GPU‐acceleration support. Different approaches have been developed regarding various aspects of the algorithms, with important differences in the specific solutions. Focusing on published works in the field of classical MD, we describe the chosen implementation methods and algorithmic techniques used for porting to GPU, as well as how recent advances of GPU architectures will provide even more optimization possibilities in the future. This article is characterized under: Software > Simulation Methods Computer and Information Science > Computer Algorithms and Programming Molecular and Statistical Mechanics > Molecular Dynamics and Monte‐Carlo Methods

show abstract

Impact of multicores on large-scale molecular dynamics simulations

Abstract: Processing nodes of the Cray XT and IBM Blue Gene Massively Parallel Processing (MPP)

Cited by 9 publications

References 10 publications

Performance analysis of a hybrid MPI/OpenMP application on multi-core clusters

Performance analysis of a hybrid MPI/OpenMP application on multi-core clusters

Multi‐scale characterization of the energy landscape of proteins with application to the C3D/Efb‐C complex

Classical molecular dynamics on graphics processing unit architectures

Contact Info

Product

Resources

About