Data Parallel Programming A Transition Sequential to Parallel Computing Janet Hartman United States Air Force Academy af.mil from students, like sorting, finding multiplying developing matrices. an algorithm the maximum element in a list, or Students can focus on the aspects of without the complexity of the problem hartman@kirk.usafa. being an impediment Dean Srutders Illinois State University sattders@ilstu.bitnet performance Parrdlel computing many computer introduced has become a popular topic in to their understanding. Students should also know how to analyze the of a parallel algorithm and compare it to its most et%cient sequential counterpart. science curricula. Usuatly students are the cost of both. to the intricacies of parallel programming computation: Vastly via one or multiplying more of three models of parallel message passing, or data parallel. required to develop efficient Currently, which parallel a parallel shared memory, execution different thinking is time of the algorithm. This can be done by examining is computed by The cost of an rdgorithm the number of processors used to execute it by the Since a sequential algorithm is always executed on one processor, programs for each of these models. execution no literature addresses the question of time of the rdgorithm. the cost is
/lp/association-for-computing-machinery/data-parallel-programming-a-transition-from-serial-to-parallel-tey4jQfROB