Search Results for "mpi"

...Its custom-built operating system (OS) bypass hardware interface enhances the performance of inter-instance communications, which is critical to scaling these applications. With EFA, High Performance Computing (HPC) applications using the Message Passing Interface (MPI) and Machine Learning (ML) applications using NVIDIA Collective Communications Library (NCCL) can scale to thousands of CPUs or GPUs. As a result, you get the application performance of on-premises HPC clusters with the on-demand elasticity and flexibility of the AWS cloud.

...The FPS measures may be used in benchmarking, e.g. while optimizing HPC algorithms for performance. progrep supports both single-threaded and parallel (multicore/multinode - e.g. OpenMP/MPI) jobs. progrep can also report for jobs running on remote hosts, e.g. jobs running on Linux Clusters. progrep works in client-server model. The server can be installed in your simulation source code with only 4 extra lines (API). When the progrep command is invoked, it queries this server as a client. progrep does not interfere with or slow down your simulation when not invoked. ...

mpiP is a lightweight profiling library for MPI applications.

ParaStation is a robust and efficient cluster middleware, consisting of a high-performance communication layer (MPI) and a sophisticated management layer. Please notice, the public development has moved to github: https://github.com/ParaStation

PARUS is a data-flow parallel programming language that allows to build parallel programs for clusters and MPP multiprocessors. The data-flow graph is automatically converted to the C++/MPI source and linked with the libparus runtime library. Also there are available tools for benchmarking cluster interconnect and visualize it.

DMTA (Distributed Multithreaded Apriori) is a parallel implementation of Apriori algorithm, which exploits the parallelism at the level of threads and processes, seeking to perform load balancing among the cores. Was implemented in C++ language, using the parallelization libraries OpenMP and MPI. The algorithm was generated as a result of a project developed by André Camilo Bolina, under the guidance of teachers Marluce Rodrigues Pereira, Ahmed Ali Abdalla Esmin and Denilson Alves Pereira, in Department of Computer Science at Federal University of Lavras. The results of this project were published in the Revista de Sistemas de Informação da FSMA and is available in http://www.fsma.edu.br/si/edicao11/FSMA_SI_2013_1_Principal_1.html

...Additionally the package includes basic stochastic simulation algorithms. The package includes several example networks which can be easily modified for other networks. Serial and parallel MPI implementations can be built and called from the command line. The serial implementation of the package can also be called from Matlab via the MEX interface with limited functionality. It supports many user specified options, including the use of antithetic sampling that can be used in some cases to increase efficiency.

This project is a tool for beginners to understand why a MPI program works or doesn't. It aims simple MPI codes, with a modest number of Processes. It is a real time view of an executing code, providing help even when the program deadlocks.

Distributed Parallel Programming for Python! This package builds on traditional Python by enabling users to write distributed, parallel programs based on MPI message passing primitives. General python objects can be messaged between processors. Ru

The STAR-MPI (Self-tuned adaptive routines for MPI collective routines) project is an on-going research and development effort to apply the adaptive software approach to develop efficient MPI collective routines across platforms and applications.

MPIUnit is a unit testing framework for MPI programming.

MorphMPI simulates an ABI for MPI. Applications compiled using MorphMPI can easily be relinked to use any MPI implementation on a compatible OS-cpu combination.

This Project implements the mergesort algorithm in a MPICH2 (parallel programming -) environment.

mib is an MPI-based I/O benchmark for clusters.

pympy offers a Python package compatible with most popular MPI (Message Passing Interface) libraries. It gives a Python program an almost complete access to the MPI standard v1.2, supporting MPI native datatypes, plus python objects and numeric arrays.

The Arbitrarily-Decomposed MPI FFT library is a C subroutine library for the computation of the multidimensional Fast Fourier Transform (FFT) of a distributed grid using the Message-Passing Interface (MPI).