Developing a High Performance Software Library with MPI and CUDA for Matrix Computations
Journal Title: Computational Methods in Social Sciences - Year 2013, Vol 1, Issue 2
Abstract
Nowadays, the paradigm of parallel computing is changing. CUDA is now a popular programming model for general purpose computations on GPUs and a great number of applications were ported to CUDA obtaining speedups of orders of magnitude comparing to optimized CPU implementations. Hybrid approaches that combine the message passing model with the shared memory model for parallel computing area solution for very large applications. We considered a heterogeneouscluster that combines the CPU and GPU computations using MPI and CUDA for developing a high performance linear algebra library. Our library deals with large linear systems solvers because they are a common problem in the fields of science and engineering. Direct methods for computing the solution of such systems can be very expensive due to high memory requirements and computational cost. An efficient alternative are iterative methods which computes only an approximation of the solution. In this paper we present an implementation of a library that uses a hybrid model of computation using MPI and CUDA implementing both direct and iterative linear systems solvers. Our library implements LU and Cholesky factorization based solvers and some of the non-stationary iterative methods using the MPI/CUDA combination. We compared the performance of our MPI/CUDA implementation with classic programs written to be run on a single CPU.
Authors and Affiliations
Bogdan Oancea, Andrei Tudorel
Keywords
Related Articles
- EP ID EP88071
- DOI -
- Views 157
- Downloads 0
M1 and M2 indicators- new proposed measures for the global accuracy of forecast intervals
This is an original scientific paper that proposes the introduction in literature of two new accuracy indicators for assessing the global accuracy of the forecast intervals. Taking into account that there are not specifi...
Nonlinear Models for Economic Forecasting Applications: An Evolutionary Discussion
This article follows the main contributions brought to the nonlinear modeling literature. We investigate and review a series of parametric initiatives, focusing on the evolution of TAR and ARCH – GARCH model families in...
Developments of regional gross domestic product and the unemployment rate in Romania during 2000-2012
Economic development of territorial units can be characterized by measuring the ability of producing goods and services and by measuring labor force in the area. The capacity of producing goods and service...
On MDA-SOA based Intercloud Interoperability framework
Cloud computing has been one of the latest technologies which assures reliable delivery of on - demand computing services over the Internet. Cloud service providers have established geographically distributed data center...
Software solutions for identifying outliers
An outlier is an observation that appears to deviate evidently from other observations in the sample. It is important to identify an outlier because it may suggest erroneous data or, in some cases, outliers may be due to...