If you're looking for the hardware structural description, well there are two ways you can implement a multiplier: (1) Serial multiplier (2) Parallel Multiplier.
Serial multiplier takes much less space compared to a parallel multiplier, but it takes more time and parallel multiplier vice-versa. It's a matter of trade-off between size & speed. Multiplication is actually a sequence of add-shift process.
So, if you're building an nxn bit multiplier, for a serial multiplier you need only 1 n-bit adder with 1 2*n-bit shift register and 1 n-bit register. You need at least 8 clock cycles for this circuit. For a parallel multiplier, you need n n-bit adder but you circuit becomes a single combinational logic block.
-lordwolf-