Publication
Rapid single-flux quantum (RSFQ) is one of the most advanced cryogenic superconductive electronics technologies. With orders of magnitude smaller power dissipation, RSFQ is an attractive technology for cloud computing, aerospace electronics, and high-speed interfacing with quantum computing systems. Technological challenges however greatly complicate the realization of VLSI-complexity RSFQ systems. For example, gate-level pipelining in SFQ systems incurs a significant area overhead due to the need for path balancing. This issue is particularly detrimental to SFQ systems due to the limited layout density of RSFQ systems.Multiple advanced SFQ logic cells exist that can be efficiently realized using SFQ technology. For example, a T1-cell can realize the full adder function with only half the area required by the conventional realization. This cell however imposes complex constraints on input signal timing, complicating its use.Multiphase clocking, recently proposed to reduce the path balancing overhead, is an effective tool for controlling the timing of the signals within a network. By utilizing multiphase clocking, the timing of the input signals of the T1-cells can be efficiently satisfied, enabling its use within the SFQ networks. In this paper, we propose a two-stage SFQ technology mapping methodology supporting the T1-cells. During the logic synthesis stage, specific parts of the SFQ network are replaced by efficient T1-cells. During the retiming stage, phases are assigned to each logic gate within the network and DFFs are inserted to satisfy the timing constraints. Using our method, the area of the SFQ networks is reduced, on average, by 6% with up to 25% reduction in optimizing the 128-bit adder.