Suppose we want to implement the instruction lwr $ra, $rb($rc) which works like a regular lw instruction but uses a register value ($rb) instead of an immediate value to calculate a memory address.

(a) The RISC approach (which MIPS takes) is to treat lwr as a pseudo-instruction and rewrite it in terms of existing MIPS instructions. Determine a minimum set of instructions to do this.

(b) The CISC approach is to modify the hardware to handle the new instruction. Assume that in doing so we must increase the clock period by 8%. What percentage of a programs instructions must be lwrs to make this approach perform better than the RISC approach? You may assume a constant CPI for all instructions.