Recall the 10-stage pipeline we studied as an example earlier in lecture. The target address of branches and jumps is ready at the end of s3. The branches are resolved at the end of s5. With the static predicted not-taken scheme, untaken branches do not have stall cycles and taken branches have 4 cycles penalty. In this problem, we study how dynamic branch predictors and BTB can improve the performance. Assume the following. Ignore jumps. Branches are 20% of all instructions executed. 60% of branches are taken. The base CPI is 1.1 without branch stalls. . a) (5 points) What is the average CPI if the processor adopts the static predicted not-taken method? b) (10 points) Suppose we implement a dynamic branch predictor. The prediction accuracy is 90% for both taken and untaken branches. The prediction is available at the end of S2. However, the processor the processor has the dynamic branch predictor