HPF and MPI Implementation of the NAS Parallel Benchmarks Supported by Integrated Program Engineering Tools
نویسندگان
چکیده
High Performance Fortran (HPF) compilers and communication libraries with the standardized Message Passing Interface (MPI) are becoming widely available, easing the development of portable parallel applications on distributed-memory parallel processor systems. The recently developed Annai tool environment supports programming, debugging and tuning of both HPFand MPI-based applications. Considering code development and subsequent maintenance time to be as important as ultimate performance, we address how sequential Fortran-77 versions of the familiar NAS Parallel Benchmark kernels can be expediently parallelized with appropriate tool support. While automatic parallelization of scientific applications written in traditional sequential languages remains largely impractical, Annai provides users with high-level language extensions and integrated program engineering support tools. In this paper, Annai support is demonstrated primarily focusing on the MG (multigrid) kernel, with complementary examples selected from the other four kernels. Respectable performance and good scalability in most cases are obtained with this straightforward parallelization strategy, even without recourse to platformspecific optimizations or major program transformations.
منابع مشابه
MPIEcho: A Framework for Transparent MPI Task Replication
In this paper we describe MPIEcho: a profiling–layer library for replicating MPI ranks independently of application parallelism. This replication effectively breaks the tight coupling between an application’s understanding of the parallel topology and that provided by the underlying MPI implementation, allowing a variety of use cases such as fault detection, independent parallelization of profi...
متن کاملImplementation of NAS Parallel Benchmarks in High Performance Fortran
We present an HPF implementation of BT, SP, LU, FT, CG and MG of the NPB2.3-serial benchmark set. The implementation is based on HPF performance model of the benchmark specific primitive operations with distributed arrays. We present profiling and performance data on SGI Origin 2000 and compare the results with NPB2.3. We discuss advantages and limitations of HPF and pghpf com-
متن کاملOn the Implementation of SPMD Applications Using Haskell
Commodities-built clusters, a low cost alternative for distributed parallel processing, brought high-performance computing to a wide range of users. Most of them are interested in parallelising scientific applications. Message passing parallel programming using lower level mechanisms, such as MPI and PVM libraries, has become popular, because these tools are free, easy to obtain, and efficient....
متن کاملThe NAS Parallel Benchmarks 2.0
We describe a set of implementations of the NAS Parallel Benchmarks based on Fortran 77 and the MPI message passing standard. These implementations, which are intended to be run with little or no tuning, approximate the performance a typical user can expect for a portable parallel program on a distributed memory computer. They complement rather than replace the original NAS Parallel Benchmarks....
متن کاملPerformance assessment of parallel techniques
The goal of this work is to evaluate and compare the computational performance of the most common parallel libraries such as Message Passing Interface (MPI), High Performance Fortran (HPF), OpenMP and DVM for further implementations. Evaluation is based on NAS Parallel benchmark suite (NPB) which includes simulated applications BT, SP, LU and kernel benchmarks FT, CG and MG. A brief introductio...
متن کامل