Jehad designed a CPU, which has an execution time for programA of $600$ seconds. Sami proposes a single cycle datapath and plans to use a compiler that will generate $300$ Billion instructions for the same program.

Jamila proposes a pipelined data path, with a clock period of $1\mathrm{.}2$ ns$.$ Her compiler will generate $400$ Billion instructions. Of these $400$ Billion instructions, $25\%$ are loads, and $20\%$ are branches. $40\%$ of all loads cause a $2-$cycle load$-$use stall. The penalty for a miss$-$ predicted branch is $5$ cycles. The processor has full bypassing and does not suffer from any other stalls. How fast does Sami need to make his clock cycle to make the program run faster than Jehad's? At least less than

a$)2$ns

b$)3$ns

c$)4$ns

d$)$ impossible

e$)$ none of the above

$18*.$ How accurate does Jamila's branch predictor need to be to make the program run faster than Jehad's? At least greater than c$)95\%$

a$)80\%$

b$)90\%$

d$)$ impossible

e$.$ none of the above