Dual-mode floating-point multiplier architectures with parallel operations

Although most modern processors have hardware support for double precision or double-extended precision floating-point multiplication, this support is inadequate for many scientific computations. This paper presents the architecture of a quadruple precision floating-point multiplier that also supports two parallel double precision multiplications. Since hardware support for quadruple precision arithmetic is expensive, a new technique is presented that requires much less hardware than a fully parallel quadruple precision multiplier. With this architecture, quadruple precision multiplication has a latency of three cycles and two parallel double precision multiplications have latencies of only two cycles. The multiplier is pipelined so that two double precision multiplications can begin every cycle or a quadruple precision multiplication can begin every other cycle. The technique used for the dual-mode quadruple precision multiplier is also applied to the design of a dual-mode double precision floating-point multiplier that performs a double precision multiplication or two single precision multiplications in parallel. Synthesis results show that the dual-mode double precision multiplier requires 43% less area than a conventional double precision multiplier. The correctness of all the multipliers presented in this paper is tested and verified through extensive simulation.