3 (25 points) Suppose you are trying to parallelize an application on multiple processor systems. The ideal speed up is the number of processors. For example, 4 times speedup with 4 processors. However, the actual speedup is much lower because of Amdahl's law (because not all the code in the application can be parallelized) and overhead (e.g., communications) introduced by parallelization. Suppose 80% of the application is parallelizable. And every time you add one processor, there is an overhead of 1% of the original execution time (of the code you parallelize). Include overhead in questions after question a) What is the speedup you can achieve with 8 processors without considering overhead? What is the best speedup you can achieve without considering overhead? What is the speedup you can achieve with 4 processors? a) b) c) What is the speedup you can achieve with 8 processors? d) What is the speedup you achieve by doubling the number of processors from 4 to 8? e) What is the best speedup you can achieve? How many processors would you use to achieve the best performance