For floating point operations
X=(mx,ex)=mx*2ex Y=(my,ey)=my*2ey
X+Y=(mx*2ex-ey+my)*2ey
Floating point operations dominant in scientific applications!