Please explain answer
1.9 (30 pts) We are running on a system with three classes of instructions: arithmetic instructions, load store instructions, and branch instructions with. Our program executes the number of instructions shown per category in the first row below. Given the IC, the CPI, and processor speed, we can calculate an execution time. What happens if we add more processors? Does having processors mean our program will run in lp the time? Not quite. We have to do extra work to coordinate things to make sure we don't step on each other. To account for this, we will multiply the number of processors by 70%. So instead of dividing up IC across p processors, it's as if we only had (p.7). Use this for math and load store commands but not for branches-that number is constant. (26 pts)
Find the total execution time of this program for each scenario shown, and show the relative speedup relative to the single processor result. Use the following table for your answers. The original number of instructions on which you will base subsequent rows is given The speed of the processor is: 6.00 GHz. Don't show more than 3 digits of significance anywhere. #arith instr CPI Arith #LS CPI #branch instr L/S instr CPI Branch cycles exec time (sec) speedup 5.00E91 2.00E9 9 1.00E9 1.00 1000 b. (4 pts) [10] To what value would the load store CPI for 1 processor have to change to match the execution time of the 4-processor system? In other words, if you want the l-processor system to run your program in the same execution time as the 4-processor system, figure out what the Load Store CPI needs to change to. In Excel, you can use a Goal Seek function to check your answer, but I want to see your equation Here's something that might help. Let's call math instructions category 1, load/store 2 and branch 3. The execution time can be calculated with: IC X CPI. + IC X CPI + ICX CPI Y GHS If you replace the execution time for the single-processor system with that for the 4-processor row t, and solve for CPIZ', you'll have your answer. What is it? 1.9 (30 pts) We are running on a system with three classes of instructions: arithmetic instructions, load store instructions, and branch instructions with. Our program executes the number of instructions shown per category in the first row below. Given the IC, the CPI, and processor speed, we can calculate an execution time. What happens if we add more processors? Does having processors mean our program will run in lp the time? Not quite. We have to do extra work to coordinate things to make sure we don't step on each other. To account for this, we will multiply the number of processors by 70%. So instead of dividing up IC across p processors, it's as if we only had (p.7). Use this for math and load store commands but not for branches-that number is constant. (26 pts) Find the total execution time of this program for each scenario shown, and show the relative speedup relative to the single processor result. Use the following table for your answers. The original number of instructions on which you will base subsequent rows is given The speed of the processor is: 6.00 GHz. Don't show more than 3 digits of significance anywhere. #arith instr CPI Arith #LS CPI #branch instr L/S instr CPI Branch cycles exec time (sec) speedup 5.00E91 2.00E9 9 1.00E9 1.00 1000 b. (4 pts) [10] To what value would the load store CPI for 1 processor have to change to match the execution time of the 4-processor system? In other words, if you want the l-processor system to run your program in the same execution time as the 4-processor system, figure out what the Load Store CPI needs to change to. In Excel, you can use a Goal Seek function to check your answer, but I want to see your equation Here's something that might help. Let's call math instructions category 1, load/store 2 and branch 3. The execution time can be calculated with: IC X CPI. + IC X CPI + ICX CPI Y GHS If you replace the execution time for the single-processor system with that for the 4-processor row t, and solve for CPIZ', you'll have your answer. What is it
