Consider an implementation of the instruction set architecture P1 has a clock rate of 4.8 GHz and CPIs (Cycles Per Instruction) of 1 for arithmetic, 4 for load/store and 8 for branch. A program execution consists of 1 million instructions, including 50% arithmetic, 30% load/store, and 20% branch instructions.
In this case, the average CPI is ________.
For a 10% improvement in performance of P1, a load/store instruction will take ______ cycles on average if other classes are NOT improved at all.