Summary: | This report describes the parallel implementation of the thinning and skeletonization problem, an image processing application from the Cowichan problem set. The problem was implemented in Orca, an Algol-like language supporting distributed shared memory, and executed on a dedicated pool of 80 SPARC processors connected by an Ethernet. The author's experiences of programming in Orca, and the results obtained for the two phases of this problem, are described. 1 Introduction Parallelism offers high performance, but programming parallel systems is difficult. One reason for this is that parallel programmers have to worry about more things than sequential programmers, such as synchronization and data consistency. For the potential of parallelism to be realized, parallel programming systems which are as easy to use as their sequential counterparts, must be developed. This report discusses an investigation of the use of Orca, a language offering distributed shared memory, to parallelize .
|