Abstract

This paper describes implementation of a 3-D finite element analysis technique on FPS-T20, a parallel processing computer with the vector processors arranged in a hypercube topology. Details of the implementation procedure, such as generation of the finite element stiffness equations using vectorized algorithms and solution of the resulting equilibrium equations using the parallel preconditioned conjugate gradient method, are presented. In order to evaluate the performance of T20, studies are conducted to determine the traditional performance parameters such as speedup and efficiency. In addition, a series of analyses are carried out in order to determine the largest size of a finite element mesh that can be accommodated on a FPS-T20. The results of the speedup and efficiency studies indicate trends which are similar to those observed previously for hypercube computers with scalar processors. It is also demonstrated that parallel/vector computing systems with hypercube topology can be used to solve large-finite element analyses.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call