Significance of parallel computation over serial computation

In today's scenario there is a need of fast computers to perform huge tasks in less time. In serial computation one task will be done after another but it takes more time. On the other hand, time taken by a computation problem can be reduced by performing several operations simultaneously. Parallel computing [4,8,9] is the concurrent use of multiple resources to solve a single problem. A computational problem is broken into different discrete parts that can be solved independently. Suppose, in a task there are n number of operations so in serial execution it will take O(n) time but if we use n processors and execute each operation in each processor then total time complexity will be O(1) which is very less as compared to the serial one. Although it will be greater than O(1) because of some factors like synchronization, fault tolerance, load distribution which also used to get involve among processors to work concurrently. This paper describes the need of parallel computation to increase the speedup and reduce the time complexity and also the internal architecture like how computers can be connected to work efficiently in parallel. This paper includes so many interconnection networks such as linear array, butterfly, de-brujin, shuffle exchange network etc. Also it describes the models, RAM model for serial computation and PRAM model for parallel computation.