Chapter 6. Adding Parallel Programming Capabilities to C Through the PVM

Chapter 6. Adding Parallel Programming Capabilities to C++ Through the PVM

"We have thus divided our problem into two parts . The child-programme and the education process. These two remain very closely connected. We cannot expect to find a good child-machine at the first attempt. One must experiment with teaching one such machine, and see how well it learns . . ."

” Alan Turing, Can A Machine Think?

In this Chapter

• The Classic Parallelism Models Supported by PVM

• The PVM Library for C++

• The Basic Mechanics of the PVM

• Accessing Standard Input (stdin) and Standard Output ( stdout ) within PVM Tasks

• Summary

The PVM (Parallel Virtual Machine) is a software system that provides the software developer with the facilities to write and run programs that exploit parallelism. The PVM presents a collection of networked computers to the developer as a single logical machine with parallel capabilities. The collection of computers can all have the same architecture or the collection can consist of computers with different architectures. The PVM can even be connected to computers that fall into the MPP (Massively Parallel Processor) class. Although PVM programs can be developed for a single computer, the real advantages come when there are two or more computers connected.

The PVM supports the message passing model as a means of communication between concurrently executing tasks. An application interacts with the PVM through a library that consists of APIs for process control, sending messages, receiving messages, signaling processes, and so on. A C++ program interfaces with the PVM library in the same way that it interacts with any other function library. While a program that accesses PVM library calls does require certain functions to be called to initialize the environment, there is nothing that forces any particular form or architecture on a C++ program. This means that the C++ programmer can combine PVM capabilities with other styles of C++ programming (e.g., object-oriented, parameterized programming, agent-oriented programming, and structured programming). The use of libraries to provide additional functionality to C++ is considered one of its advantages. Through the use of libraries such as PVM, MPI, or Linda, a C++ developer can use different models of parallelism, whereas other languages are restricted to whatever parallel primitives are built into the language. The PVM library is perhaps the easiest way to add parallel programming capabilities to the C++ language.