There exist many types of problems that benefit from parallel computing, primarily due to reasons of geographical or computational intensity. In the former case, these include data acquisition, distributed control and monitoring systems. The latter case includes matrix operations, database operations, simulations, signal processing, ray tracing, data compression and decompression, encryption and decryption. Among the research applications that use parallel computing are computer simulation to build safer cars, design of commercial aircraft that resulted in better fuel efficiency and safety, mapping of DNA genes with the hope to find cures for fatal diseases, and signal processing for weather forecasts. One ambitious project that attempts to harness the parallel computation power from all over the world is the Search for Extra-Terrestrial Intelligence (SETI) project. The SETI project analyses the cosmic rays from the universe captured from the Arecibo Radio Observatory in Puerto Rico by distributing the signal captured to millions of idling home user computers via Internet in search for extra-terrestrial origin. The SETI project is an example of a problem that cannot be easily implemented using sequential computing. The 1980's were an important decade in the development of parallel computers, with the advent of the CRAY, the ICE DAP and the Goodyear Aerospace MPP. These are among the viable systems used to solve hard 'number-crunching' tasks. A new trend in parallel computing started with the availability of cheap micro-processors and memories to form low-cost Multiple Instruction Multiple Data (MIMD) computers, cormected via a common bus to a shared memory. Data was commrrnicated between these processors in the shared memory, with the support of software. The performance of these low-cost computers could approach the performance of super-computers at a fraction of the cost. However, as more microprocessors were added into the parallel network, bus contention increased and therefore degraded the overall performance. A compromise solution was later made available, where the addition of microprocessors into parallel networks that did not increase bus contention. This solution was the Transputers family. Transputers were developed by Inmos Limited, and had many attractive features. These features include four on-chip dedicated cormmmication links and microcoded software support for scheduling and communications that does not require an OS. Transputers were used as a simple building block, to construct complex networks that could perform multiprocessing tasks, either as a distributed control system or as 'number crunching' operations. Using a large number of these components together, powerful supercomputers could be built. A transputer could essentially be run without requiring additional hardware support logic; the code developed does not need software support. Therefore developers could achieve faster prototyping, reduce the design cycle, and lower development costs. As a low cost and yet efficient alternative to other parallel solutions such as supercomputers and shared memory computers, the transputer attracted many companies, especially those specialising in embedded systems. The Transputer family could be divided into two generations, the first generation, which includes the T1, T2 and T4 series, and the later second generation, which was the T9000 series. The first generation uses the 0SLinks (over-sample link), and the second generation uses DSLinks, which were incompatible with OSLinks. OCCAM was the main language for the transputer systems, but applications were also written in Parallel C. A more recent parallel computing solution is available from the desktop PC environment. With the cost of desktop PCs declining every year, and the advancement of data communication technology (for instance Ethernet[6]which offers 10Mbits to 100Mbits per second), many researchers have enhanced the computing power of the desktop environment with the use of clusters. A cluster is a group of independent computer systems, grouped together in a network to form a parallel computer, which can, at its limit, rival supercomputers. Inter-processor communications are performed via hubs or switches, and applications are parallelised using built-in libraries like Parallel Virtual Machine (PVM) or Message Passing Interface (MPI) for message-passing mechanisms. In this thesis, a distinction is drawn between the transputer solution and the desktop PC solution, i.e. the transputer is termed as the embedded solution, and the PC is termed as the cluster solution. The project described in the thesis will concentrate primarily on parallel embedded systems solutions.