Design Recipes for FPGAs
276
but if the bit is one, add the shifted multiplicand (6 in this
case).
The way we implement this in practice is to have a ‘partial prod-
uct’ and then add the shifted multiplicand (or zeros) at each stage
of the process until the multiplication is complete.
While this approach works for unsigned binary numbers, it does
not work for twos complement numbers. In the case of twos com-
plement, using a similar approach requires the addition of sign bits
to the left of the shifted multiplicand at each stage and then a final
step of negating the multiplicand and adding the final shifted value
to the partial product. A simpler approach ...