Abstract
This paper demonstrates that scalability and competitive efficiency can be achieved for unstructured grid finite element applications on distributed memory machines, such as the Connection Machine CM-5 system. The efficiency of finite element solvers is analyzed through two applications: an implicit computational aerodynamics application and an explicit solid mechanics application. Scalability of mesh decomposition and of data mapping strategies is also discussed. Numerical examples that support the claims for problems with an excess of fourteen million variables are presented.
Highlights
Industrial design using nite element analysis software often requires solving problems involving hundreds of thousands or even several million degrees of freedom
Distributedmemory parallel systems have been shown to be a useful alternative to classical vector supercomputers when solving those large-scale problems because of their potentially greater computing power and their large memory
Scepticism sometimes arises as how nite element techniques would perform on parallel systems equipped with a large number of processors
Summary
Industrial design using nite element analysis software often requires solving problems involving hundreds of thousands or even several million degrees of freedom. Distributedmemory parallel systems have been shown to be a useful alternative to classical vector supercomputers when solving those large-scale problems because of their potentially greater computing power and their large memory. Scepticism sometimes arises as how nite element techniques would perform on parallel systems equipped with a large number of processors. We intend to show that the major components of a nite element application can use e ciently such parallel computers, regardless of the number of processors.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have