Enrique S. Quintana

Implementing openMP for clusters on top of MPI

llc is a language designed to extend OpenMP to distributed memory systems. Work in progress on the implementation of a compiler that translates llc code and targets distributed memory platforms is presented. Our approach generates code for communications directly on top of MPI. We present computational results for two different benchmark applications on a PC-cluster platform. The results reflect similar performances for the llc compiled version and an ad-hoc MPI implementation, even for applications with fine-grain parallelism.

Parallelization of the GNU Scientific Library on heterogeneous systems

In this paper we present our joint efforts towards the development of a parallel version of the GNU Scientific Library for heterogeneous systems. Two well-known operations arising in discrete mathematics and sparse linear algebra allow us to describe the design and the implementation of the library, and to report experimental results on heterogeneous clusters of personal computers.

SOLVING DISCRETE-TIME LYAPUNOV EQUATIONS FOR THE CHOLESKY FACTOR ON A SHARED MEMORY MULTIPROCESSOR

In this paper we study the parallel solution of the discrete-time Lyapunov equation. Two parallel fine and medium grain algorithms for solving dense and large order equations

Parallelization of GSL: Architecture, Interfaces, and Programming Models

\hat A\hat X\hat A^T - \hat X + \hat B\hat B^T = 0

Parallel Algorithms for Solving the Algebraic Riccati Equation via the Matrix Sign Function

on a shared memory multiprocessor are presented. They are based on Hammarling’s method and directly obtain the Cholesky factor of the solution. The parallel algorithms work following an antidiagonal wavefront. In order to improve the performance, column-block-oriented and wrap-around algorithms are used. Finally, combined fine and medium grain parallel algorithms are presented.

Parallelization of GSL: performance of case studies

In this paper we present our efforts towards the design and development of a parallel version of the Scientific Library from GNU using MPI and OpenMP. Two well-known operations arising in discrete mathematics and sparse linear algebra illustrate the architecture and interfaces of the system. Our approach, though being a general high-level proposal, achieves for these two particular examples a performance close to that obtained by an ad hoc parallel programming implementation.

Numerical solvers for discrete-time periodic riccati equations

Abstract In this paper parallel algorithms for solving the continuous-time algebraic Riccati equation both on shared memory multiprocessors and distributed memory multiprocessors are presented. The algorithms are based on the matrix Sign function of the Hamiltonian matrix associated with the equation. In the case of shared memory multiprocessors, the use of LAPACK and BLAS-3 subroutines allows the generation of reliable, portable and efficient subroutines for solving the algebraic Riccati equation. In the case of distributed memory multiprocessors, the distribution of the data wrapped by blocks of columns among the processors and the use of LAPACK and BLAS-3 subroutines improves the efficiency of the algorithms as reported in the experimental results.

Parallelization of GSL: the Web service interface

In this paper we explore the parallelization of the scientific library from GNU both on shared-memory and distributed-memory architectures. A pair of classical operations, arising in sparse linear algebra and discrete mathematics, allow us to identify the major challenges involved in this task, and to analyze the performance, benefits, and drawbacks of two different possible parallelization approaches.

AN EFFICIENT PARALLEL SYLVESTER EQUATION SOLVER BASED ON THE HESSENBERG-SCHUR METHOD∗

Abstract This paper analyzes the performance of two different methods for solving discrete-time periodic Riccati equations. The first approach is based on the computation of the periodic Schur form of the monodromy matrices, the reordering of the eigenvalues in this form, and the solution of certain linear systems. The second approach performs a sequence of orthogonal swaps in the monodromy matrices, and then employs the so-called matrix disk function, to solve the equations. Numerical experiments arts reported for both methods on serial and shared memory platforms.

Toward the parallelization of GSL

We present our joint effort to develop a Web based interface for the GNU Scientific library and its parallelization. The interface has been developed using standard Web services technology to enable the use of non local resources to execute parallel programs. The final result is a computing service where sequential and parallel routines demanding high performance computing are supplied. The design allows to incorporate new servers and platforms with a small number of software requirements.