Consider the code snippets below. In the first line of each code, replace N with $200.$

a$)$ Consider the five$-$stage pipeline seen in class, with branches resolved in the ID stage, predictor branch not taken, with all possible forwarding. How many cycles will it take to execute the Soma$1$ code snippet?

A superscalar pipeline is capable of executing multiple instructions each cycle. In a superscalar with static scheduling, the compiler is responsible for generating groups of independent instructions that can be executed together. When it is not possible to group instructions into a block $($e$.$g$.,$ due to data dependencies between instructions$),$ the compiler generates nops to complete the groups. In the next questions, consider a superscalar pipeline with static scheduling releasing two instructions per cycle.

b$)$ Considering the same assumptions as the normal pipeline, how many cycles will it take to execute the Soma$1$ code snippet in the superscalar pipeline? Note: Do not modify the code.

Loop unrolling is a technique that replicates the work performed in the body of the loop, reducing the number of iterations. For example, if the original loop fills $100$ positions of a randomly valued vector, an unrolled version could fill four positions in each iteration and run $25$ iterations. The Soma$1$Unroll code is an unrolling loop version $($factor $2)$ of the code in Soma$1.$

c$)$ Considering the same previous assumptions, how many cycles will it take to execute the Soma$1$Unroll code in the superscalar pipeline?