3. (8 points - Correctness) Performance: Bower found two processors in another computer room and in this question we will compare the performance of two processors using a certain benchmark. Processor L1 Size L1 D-Cache Miss Rate P1 4KB 6% P2 8KB 2% L1 I-Cache Miss Rate 2% 4% L2 Hit Time 0.4ns 0.9ns We are also given that accesses to the main memory takes 90ns. Additionally Processor 2 has to execute 15% more instructions than Processor 1. It is also seen that only 30% of the instructions access memory (i.e, lw and sw instructions). Profiling the code also showed us that 15% of the instructions were branch instructions. Both processors support branch prediction with 90% of the branches being predicted accurately and mis-predicts with P1 having a 2 cycle penalty and P2 having a 3 cycle penalty. Assume that we have no data hazards. (c) Estimate the MCPI of the benchmark for both the processors. (Hint: You need to consider both instruction and data memory). There is no overlap between instruction misses and data misses so all penalties are additive. (d) Estimate the BCPI (Contribution due to branches) for both the processors (e) Considering just this benchmark, which of these processors is faster? (Assume Base CPI = 1.0 for both) Hint: Use execution time 3. (8 points - Correctness) Performance: Bower found two processors in another computer room and in this question we will compare the performance of two processors using a certain benchmark. Processor L1 Size L1 D-Cache Miss Rate P1 4KB 6% P2 8KB 2% L1 I-Cache Miss Rate 2% 4% L2 Hit Time 0.4ns 0.9ns We are also given that accesses to the main memory takes 90ns. Additionally Processor 2 has to execute 15% more instructions than Processor 1. It is also seen that only 30% of the instructions access memory (i.e, lw and sw instructions). Profiling the code also showed us that 15% of the instructions were branch instructions. Both processors support branch prediction with 90% of the branches being predicted accurately and mis-predicts with P1 having a 2 cycle penalty and P2 having a 3 cycle penalty. Assume that we have no data hazards. (c) Estimate the MCPI of the benchmark for both the processors. (Hint: You need to consider both instruction and data memory). There is no overlap between instruction misses and data misses so all penalties are additive. (d) Estimate the BCPI (Contribution due to branches) for both the processors (e) Considering just this benchmark, which of these processors is faster? (Assume Base CPI = 1.0 for both) Hint: Use execution time