Optimized design of a double-precision floating-point multiply-add-dused unit for data dependence | IEEE Conference Publication | IEEE Xplore