The LPARX Programming System


LPARX is a C++ class library for for implementing portable scientific applications on distributed memory parallel computer architectures, including Workstation Clusters. Software can be developed on a single processor workstation, which facilitates debugging.

LPARX is intended for particle methods and multi-level adaptive finite different methods. However, it may  be used for single mesh finite difference methods as well.

LPARX isolates parallel coordination activities from numerical computation, and existing serial numerical kernels can often be utilized in LPARX applications with little change. The user may customize load balancing activity to the application, and treat interprocessor communication at a very high level.

Applications may be written in a mixture of languages and are portable across a variety of machines including the Intel Paragon, Cray C-90, IBM SP-2, networks of workstations under PVM, Cray T3D(soon) and single processor workstations. The LPARX software is implemented as a C++ class library. Those Paragon sites without C++ support should download the paragon-gnu.tar script, which installs the GNU g++ compiler on the Paragon.


Documentation


Source Distribution


LPARX Authors

12/06/00 Scott B. Baden / UCSD / baden@cs.ucsd.edu