Assume a workload that 20% of the total number of operations are floating point operations. On average, floating point operations take 8 cycles (i.e., CPI for floating point operations is 8). The rest of the instructions take 1.5 cycles, on average (i.e., CPI is 1.5 for the rest of the instructions). Among the floating point operations, there is one particular operation that takes 20 cycles (let's call it fpX), and it accounts for 5% of the overall number of instructions. Assume that we came up with two different optimizations to improve the performance. The first optimization reduces the CPI of the fpX (which is a floating point operation that takes 20 cycles) to 5. The second optimization reduces the average CPI of all floating point operations to 3. Calculate the overall CPI under two optimizations and find which optimization is better. Assume a workload that 20% of the total number of operations are floating point operations. On average, floating point operations take 8 cycles (i.e., CPI for floating point operations is 8). The rest of the instructions take 1.5 cycles, on average (i.e., CPI is 1.5 for the rest of the instructions). Among the floating point operations, there is one particular operation that takes 20 cycles (let's call it fpX), and it accounts for 5% of the overall number of instructions. Assume that we came up with two different optimizations to improve the performance. The first optimization reduces the CPI of the fpX (which is a floating point operation that takes 20 cycles) to 5. The second optimization reduces the average CPI of all floating point operations to 3. Calculate the overall CPI under two optimizations and find which optimization is better