> For example, multiply-and-add is usually one cycle in float, but would always be two separate instructions in integers.

Unless the multiply is by a constant 2, 4, or 8, in which case you can use `lea`.