A discussion is provided of the use of parallel processing clusters and PVM (Parallel Virtual Machine) for network cluster supporting UNIX and Linux. Beowulf can also be used for clustering workstations together for parallel processing, but the advantage of PVM over conventional approaches is that it is freely available, needs no special hardware, is portable, and supports many UNIX and Linux platforms. Important PVM features include heterogeneous support for combining workstations of different architectures , scalability, dynamic configurability, and support for the SPMD and MPMD parallel processing models. The workings of PVM are described, along with installation, use, the PVM application programming interface (API), and test programs and results. Techniques and code are provided that facilitate creation of a cluster of networked workstations that can be used for parallel processing. The cluster operates as a single virtual machine to shorten elapsed execution time of multiple tasks. The test programs show that multiple tasks execute concurrently on multiple hosts. Future work with PVM would involve efficient loading of software changes over the cluster. A main use of PVM in the future could be implementation and testing of algorithms decomposed for parallel operation. Among potential algorithms would be matrix multiplication, sorting, and puzzle solutions using trees.