Parallelism is often considered the solution to the problem of need for ever more processing power. Critics cite Amdahl’s Law—the speed of an algorithm is dominated by its sequential parts—to disparage the potential of massive parallelism. It is often overlooked, however, that the law applies to algorithms, not problems. Indeed, for a given problem, no known procedure can find all applicable algorithms. Thus, while it can be proved that there are inherently serial algorithms or procedures, there is no known proof that any problems are inherently serial.
KeywordsAssociative Memory Associative Model Massive Parallelism Instruction Stream Associative Search
Unable to display preview. Download preview PDF.