Compiling SVM-Fortran for the Intel Paragon XP/S

The Early Evaluation of the Intel Paragon XP/S 5 and XP/S 35: an experimental log

Benchmarking UHGROMOS

Porting of the parallel Fortran preprocessor, Pfortran, to Intel Corporation and IBM Corporation massively parallel processor machines is presented. The machines include the Intel iPSC/SSO, the Caltech Intel DELTA, the Intel Paragon, and the IBM SPl. The UHGROMOS molecular dynamics program, a version of GROMOS ported to parallel machines through the use of Pfortran, was used as a test applicati...

Passing Programming on a Finite Element Application

This paper describes the methods used and experiences made with implementing a nite element application on three diierent parallel computers with either message passing or shared virtual memory as the programming model. Designing a parallel nite element application using message-passing requires to nd a data domain decomposition to map data into the local memory of the processors. Since data ac...

Shared Virtual Memory and Message Passing Programming on a Finite Element Application

This paper describes the methods used and experiences made with implementing a nite element application on three diierent parallel computers with either message passing or shared virtual memory as the programming model. Designing a parallel nite element application using message-passing requires to nd a data domain decomposition to map data into the local memory of the processors. Since data ac...

On Experiments with a Parallel Direct Solver for Diagonally Dominant Banded Linear Systems

We report on numerical experiments that we conducted with a direct algorithm, the single width sparator algorithm, to solve diagonally dominant banded linear systems. With detailed estimations of computation and communication cost we quantitatively analyze their innuence on the parallel performance of the algorithm. We report on numerical experiments executed on an Intel Paragon XP/S-22MP.