Next: Examples Up: Parallel Computing Previous: Virtual Shared Memory Computers

Parallel Programming Models

There are two basic types of parallel applications - single instruction multiple data (SIMD) and multiple instruction multiple data (MIMD). While in the former case the same code is working on different (and distributed) data, the latter one is based on several codes working on different data. The parallelization in SIMD model can be based on standards as Open MP or HPF (High Performance Fortran). However, their application is limited to shared memory multiprocessors and their efficiency is generally not high (depending on the underlying problem and data distribution). On the other hand, the parallelization of MIMD models is based on message passing standards as MPI or PVM. The efficiency of these applications is dependent on the actually adopted parallel computing paradigm. The advantage of the MIMD model is that it works on shared memory computers as well as on massively parallel platforms.

Next: Examples Up: Parallel Computing Previous: Virtual Shared Memory Computers

Daniel Rypl
2005-12-03