Consider the following sequence of floating$-$point instructions:

MUL.D F$1,$ F$5,$ F$7$

ADD.D F$3,$ F$1,$ F$10$

We studied several different systems for implementation of floating$-$point instructions, but for all of them, the ADD.D must wait to issue $($stalling the pipeline in some systems, and simply waiting in a data structure for others$),$ pending completion of the multiplication; also, the result of the multiplication must be passed to the adder hardware.