Comment by nand2mario

The memory pipeline just starts one cycle later than now. Effective address is calculated during the first cycle of the instruction. The microcode then waits for it to finish with the DLY (delay) micro-op, which releases one cycle later.