A novel architecture for floating-point multiply-add-fused operation | IEEE Conference Publication | IEEE Xplore