MPI Code Research Articles

A highly optimized skeleton for unbalanced and deep divide-and-conquer algorithms on multi-core clusters

AbstractEfficiently implementing the divide-and-conquer pattern of parallelism in distributed memory systems is very relevant, given its ubiquity, and difficult, given its recursive nature and the need to exchange tasks and data among the processors. This task is noticeably further complicated in the presence of multi-core systems, where hybrid parallelism must be exploited to attain the best performance, and when unbalanced and deep workloads are considered, as additional measures must be taken to load balance and avoid deep recursion problems. In this manuscript a parallel skeleton that fulfills all these requirements while providing high levels of usability is presented. In fact, the evaluation shows that our proposal is on average 415.32% faster than MPI codes and 229.18% faster than MPI + OpenMP benchmarks, while offering an average improvement in the programmability metrics of 131.04% over MPI alternatives and 155.18% over MPI + OpenMP solutions.

The Pencil Code, a modular MPI code for partial differential equations and particles: multipurpose and multiuser-maintained

The Pencil Code is a highly modular physics-oriented simulation code that can be adapted to a wide range of applications. It is primarily designed to solve partial differential equations (PDEs) of compressible hydrodynamics and has lots of add-ons ranging from astrophysical magnetohydrodynamics (MHD) to meteorological cloud microphysics and engineering applications in combustion. Nevertheless, the framework is general and can also be applied to situations not related to hydrodynamics or even PDEs, for example when just the message passing interface or input/output strategies of the code are to be used. The code can also evolve Lagrangian (inertial and noninertial) particles, their coagulation and condensation, as well as their interaction with the fluid.

Visualization Techniques of Differentially Heated Lid-Driven Square Cavity

In the present study, an attempt is made to explore the flow field inside the differentially heated lid-driven square cavity. The governing equations along with boundary conditions are solved numerically. The simulated results (100 ≤ Re ≤ 1000 and 0.001 ≤ Ri ≤ 10) are validated with previous results in the literature. The convection differencing schemes, namely, UPWIND, QUICK, SUPERBEE, and SFCD, are discussed and are used to simulate the flow using the MPI code. It is observed that the computational cost for all the differencing schemes get reduced tremendously when the MPI code is implemented. Plots demonstrate the influences of Re and Ri in terms of the contours of the fluid streamlines, isotherms, energy streamlines, and field synergy principle.

Analysis of new visualization techniques of left heated lid‐driven square cavity

AbstractIn the present study, an attempt is made to explore the flow visualization techniques inside the left‐heated lid‐driven square cavity. The governing equations along with boundary conditions are solved numerically. The simulated results (100 ≤ Re ≤ 1000 and 0.001 ≤ Ri ≤ 10) are validated with previous results in the literature. The convection differencing schemes, namely, UPWIND, QUICK, SUPERBEE, and self‐filtered central differencing are discussed and are used to simulate the flow using MPI code. It is observed that the computational cost for all the differencing schemes get reduced tremendously when the MPI code is implemented. Plots demonstrating the influences of Re and Ri in terms of the contours of the fluid streamlines, isotherms, energy streamlines, and field synergy principle are presented.

ASPECT RATIO EFFECTS ON BOTTOM HEATED 2D CAVITY USING ENERGY STREAMLINES AND FIELD SYNERGY PRINCIPLE

In the present work free convective air flow in the two-dimensional cavity with three different aspect ratios (AR) are investigated using direct numerical simulation. The bottom wall is assumed to be kept at a uniform higher temperature than that of the top wall and the other two vertical walls are assumed to be thermally insulated. The computations are conducted for Rayleigh number (Ra) values from 103 to 106. Convective schemes are compared and Self Filtered Central Differencing Scheme is used to discretize convective term. Parallel computing MPI code is adapted to run the simulations. An attempt has been made to gather the visualization techniques such as streamlines, isotherms, energy streamlines and field synergy principle to analyse the flow behaviour inside the cavity. When Ra is small, the vertical energy streamlines are observed in the cavity. As Ra further increased, the free energy streamlines observed at the boundary and the trapped energy streamlines at the centre in the horizontal direction. For a fixed Ra, and increasing AR, the average synergy angle increases. This indicates synergy or the coordination between velocity magnitude and temperature field gets decreased and leads to the growth of heat transfer rate. The field synergy principle implies by enhancing the synergy between the velocity vector and temperature gradient.

Simulation of Hypersonic-Shock-Wave–Laminar-Boundary-Layer Interaction over Blunt Fin

Shock-wave–boundary-layer interaction is one of the most important phenomena in hypersonic flights due to its ubiquitous presence and its extreme effects on the aerothermodynamic loads on the aircraft surfaces. Low-frequency oscillations in separated flows in addition to high localized pressure and heat transfer as a result of such interactions make the ability to predict the flow with shock-wave–boundary-layer interaction highly essential in design of any hypersonic aircraft. To further investigate three-dimensional shock-wave–hypersonic-boundary-layer interaction, a cylindrically blunt fin–plate junction with three different sweep angles, 0, 22.5, and 45 deg, of the fin relative to the plate is chosen to numerically simulate. An MPI code written by the authors was used to solve the unsteady Navier–Stokes and ideal gas equations. The blunt fin shock wave with strength depending on the sweep angle of the fin interacts with the boundary layer on the plate and creates a separated region with a complex flow structure. The extent of the separated region decreases with increase in the sweep angle of the fin and decrease in the strength of the blunt fin shock wave. Localized high pressure forms at the reattachment point of the boundary layer by formation of a reattachment shock wave, with its maximum peak decreasing by increase in the sweep angle of the fin. The boundary-layer thickness on the blunt fin reaches its minimum at the reattachment point, where the heat transfer rate reaches its maximum value. Unsteady oscillations are observed in the 0 deg sweep angle configuration. The unsteady oscillations significantly subside with increase in the sweep angle of the fin. Time-averaged normalized heat transfer of the numerical simulations shows good agreement with the experimental data, except for the peak heat transfer for the 0 deg sweep angle case.

Analysis Of Field Synergy In Bottom Heated Lid Driven Cubical Cavity

This study presents an innovative visualization tool for the analysis of the mixed convection in a lid-driven air filled cubical cavity heated from below. The total energy of the flow in the cavity isvisualized based on the energy stream functions or energy streamlines. Also the heat transfer enhancement in the cavity is presented with an analogy between conduction and convection, namely, the field synergy principle. Flow is assumed to be driven by the vertical temperature gradient and by the top lid of the cavity, which is assumed to slide on its own plane at a uniform speed. The top and bottom walls are assumed to be isothermal and all other walls are thermally insulated. Non dimensional governing equations of this problem are solved by using the finite volume method. Established open source CFD package OpenFOAM is utilized to investigate the flow with respect to the control parameters arising in the system. The nonlinear terms arising in the governing equations are discretized with the NVD schemes. The convection differencing schemes namely, UPWIND, QUICK, SUPERBEE and SFCD discussed and are used to simulate the flow using MPI code. It is observed that the computational cost for all the differencing schemes get reduced tremendously when the MPI code is implemented. Also SFCD scheme gave the Nuseelt number values close to those available in the literature. Extensive numerical flow visualization is conducted for the Reynolds number (Re = 100, 400, 1000) and the Richardson number (Ri = 0.001, 1, 10), which categorize the free and forced convective flow, respectively. It is observed that for a fixed value of Re, as Ri increases, the average Nusselt number (Nu¯), decreases. This shows that the natural convection starts to prevail with an increasing of Ri. But, for a fixed Ri, as Re increases (Nu¯) increases and the forced convection mode becomes dominant, leading to a chaotic flow. Plots demonstrating the influences of Re and Ri in termsof the contours of the fluid streamlines, isotherms, vortex corelines, and field synergy principle. The synergy angle of buoyant-aiding flow is high while the buoyant-opposing flow is significantly less than that of forced convection flow.

Alchemist: An Apache Spark ⇔ MPI interface

SummaryThe Apache Spark framework for distributed computation is popular in the data analytics community due to its ease of use, but its MapReduce‐style programming model can incur significant overheads when performing computations that do not map directly onto this model. One way to mitigate these costs is to off‐load computations onto MPI codes. In recent work, we introduced Alchemist, a system for the analysis of large‐scale data sets. Alchemist calls MPI‐based libraries from within Spark applications, and it has minimal coding, communication, and memory overheads. In particular, Alchemist allows users to retain the productivity benefits of working within the Spark software ecosystem without sacrificing performance efficiency in linear algebra, machine learning, and other related computations. In this paper, we discuss the motivation behind the development of Alchemist, and we provide a detailed overview of its design and usage. We also demonstrate the efficiency of our approach on medium‐to‐large data sets, using some standard linear algebra operations, namely, matrix multiplication and the truncated singular value decomposition of a dense matrix, and we compare the performance of Spark with that of Spark+Alchemist. These computations are run on the NERSC supercomputer Cori Phase 1, a Cray XC40.

Application Productivity and Performance Evaluation of Transparent Locality-aware One-sided Communication Primitives

Nowadays, the individual nodes of a distributed parallel computer consist of multi- or many-core processors allowing to execute more than one process per node. The large difference in communication speed within a node through shared memory, versus across nodes through the network interconnect, requires to use locality-aware communication schemes for any efficient distributed application. However, writing an efficient locality-aware MPI code is complex and error-prone, because the developer has to use very different APIs for communication operations within and across nodes, respectively, and manage inter-process synchronization. In this paper, we analyze and enhance a recent one-sided communication model, namely DART-MPI, which is implemented on top of MPI-3. In this runtime system, the complexities of handling locality-awareness of MPI memory access operations, either remote or local, and the related synchronization calls are hidden inside the related DART-MPI interfaces resulting in concise code and improved application and developer productivity. We have carried out in-depth evaluation of our DART-MPI system. Foremost, a micro benchmark is conducted to help understanding the prime performance overhead of implementing APIs in DART-MPI system, which is small and becomes negligible with the growing message sizes. We then compare the performance of DART-MPI and flat MPI without locality awareness, in particular blocking and non-blocking memory operations, using a realistic scientific application on a large-scale supercomputer. The comparison demonstrates that in most cases the DART-MPI version of this application shows better performance than the flat MPI version. Further, we compare the DART-MPI version to a functionally equivalent MPI version, which thus includes code to deal with data-locality, and show that DART-MPI realizes almost the full potential of highly optimized MPI while maintaining high productivity for non-expert programmers.

Comparing Coarray Fortran (CAF) with MPI for several structured mesh PDE applications

Language-based approaches to parallelism have been incorporated into the Fortran standard. These Fortran extensions go under the name of Coarray Fortran (CAF) and full-featured compilers that support CAF have become available from Cray and Intel; the GNU implementation is expected in 2015. CAF combines elegance of expression with simplicity of implementation to yield an efficient parallel programming language. Elegance of expression results in very compact parallel code. The existence of a standard helps with portability and maintainability. CAF was designed to excel at one-sided communication and similar functions that support one-sided communication are also available in the recent MPI-3 standard. One-sided communication is expected to be very valuable for structured mesh applications involving partial differential equations, amongst other possible applications. This paper focuses on a comparison of CAF and MPI for a few very useful applications areas that are routinely used for solving partial differential equations on structured meshes. The three specific areas are Fast Fourier Techniques, Computational Fluid Dynamics, and Multigrid Methods.For each of those applications areas, we have developed optimized CAF code and optimized MPI code that is based on the one-sided messaging capabilities of MPI-3. Weak scalability studies that compare CAF and MPI-3 are presented on up to 65,536 processors. Both paradigms scale well, showing that they are well-suited for Petascale-class applications. Some of the applications shown (like Fast Fourier Techniques and Computational Fluid Dynamics) require large, coarse-grained messaging. Such applications emphasize high bandwidth. Our other application (Multigrid Methods) uses pointwise smoothers which require a large amount of fine-grained messaging. In such applications, a premium is placed on low latency. Our studies show that both CAF and MPI-3 offer the twin advantages of high bandwidth and low latency for messages of all sizes. Even for large numbers of processors, CAF either draws level with MPI-3 or shows a slight advantage over MPI-3. Both CAF and MPI-3 are shown to provide substantial advantages over MPI-2.In addition to the weak scalability studies, we also catalogue some of the best-usage strategies that we have found for our successful implementations of one-sided messaging in CAF and MPI-3. We show that CAF code is of course much easier to write and maintain, and the simpler syntax makes the parallelism easier to understand.

Turbulent Transport Processes at Rough Surfaces with Geophysical Applications

In this numerical study we use the Lattice Boltzmann Method to investigate the effects of periodic and randomly rough surfaces on the turbulent transport of momentum and heat. A two-dimensional MPI code has been developed and validated against many test cases. We examine in some detail Rayleigh-Benárd convection in the presence of rough walls and study the effects of the wavelength and amplitude of the roughness distribution on the flow.

Distributed multiscale computing with MUSCLE 2, the Multiscale Coupling Library and Environment

We present the Multiscale Coupling Library and Environment: MUSCLE 2. This multiscale component-based execution environment has a simple to use Java, C++, C, Python and Fortran API, compatible with MPI, OpenMP and threading codes. We demonstrate its local and distributed computing capabilities and compare its performance to MUSCLE 1, file copy, MPI, MPWide, and GridFTP. The local throughput of MPI is about two times higher, so very tightly coupled code should use MPI as a single submodel of MUSCLE 2; the distributed performance of GridFTP is lower, especially for small messages. We test the performance of a canal system model with MUSCLE 2, where it introduces an overhead as small as 5% compared to MPI.

Performance Improvements for a Large-scale Geological Simulation

Geological models have been successfully used to identify and study geothermal energy resources. Many computer simulations based on these models are data-intensive applications. Large-scale geological simulations require high performance computing (HPC) techniques to run within reasonable time constraints and performance levels. One research area that can benefit greatly from HPC techniques is the modeling of heat flow beneath the Earth's surface. This paper describes the application of HPC techniques to increase the scale of research with a well-established geological model. Recently, a serial C++ application based on this geological model was ported to a parallel HPC applications using MPI. An area of focus was to increase the performance of the MPI version to enable state or regional scale simulations using large numbers of processors. First, synchronous communications among MPI processes was replaced by overlapping communication and computation (asynchronous communication). Asynchronous communication improved performance over synchronous communications by averages of 28% using 56 cores in one environment and 46% using 56 cores in another. Second, an approach for load balancing involving repartitioning the data at the start of the program resulted in runtime performance improvements of 32% using 48 cores in the first environment and 14% using 24 cores in the second when compared to the asynchronous version. An additional feature, modeling of erosion, was also added to the MPI code base. The performance improvement techniques under erosion were less effective.

Parallel Brownian dynamics simulations with the message-passing and PGAS programming models

The simulation of particle dynamics is among the most important mechanisms to study the behavior of molecules in a medium under specific conditions of temperature and density. Several models can be used to compute efficiently the forces that act on each particle, and also the interactions between them. This work presents the design and implementation of a parallel simulation code for the Brownian motion of particles in a fluid. Two different parallelization approaches have been followed: (1) using traditional distributed memory message-passing programming with MPI, and (2) using the Partitioned Global Address Space (PGAS) programming model, oriented towards hybrid shared/distributed memory systems, with the Unified Parallel C (UPC) language. Different techniques for domain decomposition and work distribution are analyzed in terms of efficiency and programmability, in order to select the most suitable strategy. Performance results on a supercomputer using up to 2048 cores are also presented for both MPI and UPC codes.

Parallelism Recognition Technology Based on Nested Loops Classifying

PDF HTML阅读 XML下载 导出引用 引用提醒 基于嵌套循环分类的并行识别技术 DOI: 10.3724/SP.J.1001.2012.04178 作者: 作者单位: 作者简介: 通讯作者: 中图分类号: 基金项目: “核高基”国家科技重大专项(2009ZX01036-001-001-2) Parallelism Recognition Technology Based on Nested Loops Classifying Author: Affiliation: Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:传统的分布存储并行编译系统大多是在共享存储并行编译系统的基础上开发的.共享存储并行编译系统的并行识别技术适合 OpenMP 代码生成,实现方式是将所有嵌套循环都按照相同的识别方法进行处理,用于分布存储并行编译系统必然会导致无法高效发掘程序的并行性.分布存储并行编译系统应根据嵌套循环结构的特点进行分类处理,提出适合 MPI 代码生成的并行识别技术.为解决上述问题,根据嵌套循环的结构和 MPI 并行程序的特点,提出了一种新的嵌套循环分类方法,并针对不同的嵌套循环分别提出了相应的并行识别技术.实验结果表明,与采用传统并行识别技术的分布存储并行编译系统相比,按照所提方法对嵌套循环进行分类,采用相应并行识别技术的编译系统能够更高效地识别基准程序中的并行循环,自动生成的 MPI 并行代码其性能加速比提高了 20%以上. Abstract:Existing distributed memory parallelizing compiler systems are mostly developed based on sharedsystems. The parallelism recognition technologies of shared memory parallelizing compiler systems are suitable forOpenMP code generation. Their implementation is used to recognize all nested loops by the same technology, sothat the parallelism cannot be efficiently explored when applying them to distributed memory parallelizing compilersystems. Thus, this paper proposes some parallelism recognition technologies suitable for the MPI code generationfor distributed memory parallelizing compiler systems by classifying the nested loops according to their structures.To solve these problems, a new classification method of nested loops is proposed, according to the structure ofnested loops and characteristics of MPI parallel program. Corresponding parallelism recognition technologies fordifferent nested loops are also presented, respectively. The experimental results show that compared with thedistributed memory parallelizing compiler systems that used existing parallelism recognition technologies, thecompiler systems, which use the proposed classification method and the corresponding recognition technologies,can more efficiently recognize parallel nested loops in the benchmark programs, and the performance speedup of theMPI codes automatically increased to more than 20%. 参考文献 相似文献 引证文献

Fortran code for SU(3) lattice gauge theory with and without MPI checkerboard parallelization

We document plain Fortran and Fortran MPI checkerboard code for Markov chain Monte Carlo simulations of pure SU(3) lattice gauge theory with the Wilson action in D dimensions. The Fortran code uses periodic boundary conditions and is suitable for pedagogical purposes and small scale simulations. For the Fortran MPI code two geometries are covered: the usual torus with periodic boundary conditions and the double-layered torus as defined in the paper. Parallel computing is performed on checkerboards of sublattices, which partition the full lattice in one, two, and so on, up to D directions (depending on the parameters set). For updating, the Cabibbo–Marinari heatbath algorithm is used. We present validations and test runs of the code. Performance is reported for a number of currently used Fortran compilers and, when applicable, MPI versions. For the parallelized code, performance is studied as a function of the number of processors.Program summaryProgram title: STMC2LSU3MPICatalogue identifier: AEMJ_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AEMJ_v1_0.htmlProgram obtainable from: CPC Program Library, Queen’s University, Belfast, N. IrelandLicensing provisions: Standard CPC licence, http://cpc.cs.qub.ac.uk/licence/licence.htmlNo. of lines in distributed program, including test data, etc.: 26666No. of bytes in distributed program, including test data, etc.: 233126Distribution format: tar.gzProgramming language: Fortran 77 compatible with the use of Fortran 90/95 compilers, in part with MPI extensions.Computer: Any capable of compiling and executing Fortran 77 or Fortran 90/95, when needed with MPI extensions.Operating system:1.Red Hat Enterprise Linux Server 6.1 with OpenMPI + pgf77 11.8-0,2.Centos 5.3 with OpenMPI + gfortran 4.1.2,3.Cray XT4 with MPICH2 + pgf90 11.2-0.Has the code been vectorised or parallelized?: Yes, parallelized using MPI extensions. Number of processors used: 2 to 11664RAM: 200 Mega bytes per process.Classification: 11.5.Nature of problem:Physics of pure SU(3) Quantum Field Theory (QFT). This is relevant for our understanding of Quantum Chromodynamics (QCD). It includes the glueball spectrum, topological properties and the deconfining phase transition of pure SU(3) QFT. For instance, Relativistic Heavy Ion Collision (RHIC) experiments at the Brookhaven National Laboratory provide evidence that quarks confined in hadrons undergo at high enough temperature and pressure a transition into a Quark-Gluon Plasma (QGP). Investigations of its thermodynamics in pure SU(3) QFT are of interest.Solution method:Markov Chain Monte Carlo (MCMC) simulations of SU(3) Lattice Gauge Theory (LGT) with the Wilson action. This is a regularization of pure SU(3) QFT on a hypercubic lattice, which allows approaching the continuum SU(3) QFT by means of Finite Size Scaling (FSS) studies. Specifically, we provide updating routines for the Cabibbo-Marinari heatbath with and without checkerboard parallelization. While the first is suitable for pedagogical purposes and small scale projects, the latter allows for efficient parallel processing. Targetting the geometry of RHIC experiments, we have implemented a Double-Layered Torus (DLT) lattice geometry, which has previously not been used in LGT MCMC simulations and enables inside and outside layers at distinct temperatures, the lower-temperature layer acting as the outside boundary for the higher-temperature layer, where the deconfinement transition goes on.Restrictions: The checkerboard partition of the lattice makes the development of measurement programs more tedious than is the case for an unpartitioned lattice. Presently, only one measurement routine for Polyakov loops is provided.Unusual features:We provide three different versions for the send/receive function of the MPI library, which work for different operating system +compiler +MPI combinations. This involves activating the correct row in the last three rows of our latmpi.par parameter file. The underlying reason is distinct buffer conventions.Running time:For a typical run using an Intel i7 processor, it takes (1.8-6) E-06 seconds to update one link of the lattice, depending on the compiler used. For example, if we do a simulation on a small (4 * 83) DLT lattice with a statistics of 221 sweeps (i.e., update the two lattice layers of 4 * (4 * 83) links each 221 times), the total CPU time needed can be2 * 4 * (4 * 83) * 221 * 3 E-06 seconds = 1.7 minutes,where2— two layers of lattice4— four dimensions83 * 4— lattice size221— sweeps of updating6 E-06 s— average time to update one link variable. If we divide the job into 8 parallel processes, then the real time is (for negligible communication overhead) 1.7 mins / 8 = 0.2 mins.

Hands-on Performance Tuning of 3D Finite Difference Earthquake Simulation on GPU Fermi Chipset

3D simulation of earthquake ground motion is one of the most challenging computational problems in science. The emergence of graphic processing units (GPU) as an effective alternative to traditional general purpose processors has become increasingly capable in terms of accelerating scientific computing research. In this paper, we describe our experiences in porting AWP-ODC, a 3D finite difference seismic wave propagation code, to the latest GPU Fermi chipset. We completely rewrote this Fortran-based 13-point asymmetric stencil computation code in C and MPI-CUDA in order to take advantage of the powerful GPU computing capabilities. Our new CUDA code implemented the asymmetric 3D stencil on Fermi to make the best use of GPU on-chip memory for an aggressive parallel efficiency. Benchmark on NVIDIA Tesla M2090 demonstrated 10x speedup versus the original fully optimized AWP-ODC FORTRAN MPI code running on a single Intel Nehalem 2.4GHz CPU socket (4 cores/CPU), and 15x speedup versus the same MPI code running on a single AMD Istanbul 2.6GHz CPU socket (6 cores/CPU). Sustained single-GPU performance of 143.8 GFLOPS in single precision is benchmarked for the testing case of 128x128x960 mesh size.

Towards a three-dimensional moving body incompressible flow solver with a linear deformable model

In this study, a three-dimensional fluid–structured parallelized solver is extended from the previous work (Niu et al., 2009 [1]) for moving body simulations. Based on the unified Eulerian and Lagrangian coordinate transformations, the unsteady three-dimensional incompressible Navier–Stokes equations with artificial compressibility (Chorin, 1967 [2]) in a dual-time stepping approach are first derived. To implement unsteady flow calculations, the dual-time stepping strategy including the LU decomposition method is used in the pseudo-time iteration and the second-order accurate backward difference is adopted to discretize the unsteady flow terms. Also, a third-order Roe type flux limited splitting is derived to evaluate the spatial difference of the convective fluxes. The original FORTRAN code is converted to the MPI code and tested on a 64-CPU IBM SP2. The parallel strategy here is based on the partitions of all do-loops in the original FORTRAN code and transferring the calculations inside the do-loop into different CPUs. The partition of the do-loop can be applied on the innermost loop, only or the last two inner loops depending on two-dimensional or three-dimensional problems. This kind of the parallel data partition of the loops is independent of what kind of the explicit or implicit type numerical algorithm used. Therefore, the current parallel approach can take advantage of the MPI language fully to transfer data efficiently among CPUs even for solving the governing equation implicitly. The test results show that a significant reduction of computing time in running the model and a near-linear speed up rate is achieved up to 32 CPUs at IBM SP2. The speed up rate is as high as 31 for using 64 IBM SP2 processors The test shows efficient parallel processing to provide prompt simulation of 3D cavity, unsteady dropping airfoil and blood flows in an aortic tube with a linear elastic modeling of wall motion is included here.

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

The 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.

Hybrid Message-Passing and Shared-Memory Programming in a Molecular Dynamics Application On Multicore Clusters

Hybrid programming, whereby shared-memory and message-passing programming techniques are combined within a single parallel application, has often been discussed as a method for increasing code performance on clusters of symmetric multiprocessors (SMPs). This paper examines whether the hybrid model brings any performance benefits for clusters based on multicore processors. A molecular dynamics application has been parallelized using both MPI and hybrid MPI/OpenMP programming models. The performance of this application has been examined on two high-end multicore clusters using both Infiniband and Gigabit Ethernet interconnects. The hybrid model has been found to perform well on the higher-latency Gigabit Ethernet connection, but offers no performance benefit on low-latency Infiniband interconnects. The changes in performance are attributed to the differing communication profiles of the hybrid and MPI codes.