In this article, we propose a novel architecture for single-flux-quantum (SFQ) circuits. It is compatible to the existing synchronous and asynchronous timing schemes and reduces the hardware cost for implementing the pipeline blocking and flushing, which are the two methods to improve the performance of SFQ microprocessors. We simulate two crucial elements, and the functions of the elements are verified successfully. The proposed architecture simplifies the logic design of the circuits, and it can be considered as a general architecture for SFQ circuits because of the compelling characteristics.
In this article, we propose a novel architecture for single-flux-quantum (SFQ) circuits. It is compatible to the existing synchronous and asynchronous timing schemes and reduces the hardware cost for implementing the pipeline blocking and flushing, which are the two methods to improve the performance of SFQ microprocessors. We simulate two crucial elements, and the functions of the elements are verified successfully. The proposed architecture simplifies the logic design of the circuits, and it can be considered as a general architecture for SFQ circuits because of the compelling characteristics.
Superconducting computing using single flux quantum (SFQ) technology has been recognized as a promising post-Moore's law era technology thanks to its extremely low power and high performance.Therefore, many researchers have proposed various SFQbased circuits (e.g., ALU, register file) and architectures (e.g., NPU, CPU) to exploit the potential.However, due to the absence of a reliable and high-frequency clocking scheme, general SFQ circuits cannot operate at high frequencies, making all architectural efforts for high-performance SFQ computing ineffective.In this paper, we propose SuperSFQ, a new design methodology for SFQ hardware that unlocks the high-frequency potential of SFQ technology by co-designing the clocking scheme, circuitry, and architecture.First, we propose SuperClocking, a new clocking scheme that enables high frequency in general SFQ hardware.Second, we implement an SFQ-based synchronizer to realize the reliable operation of SuperClocking.Finally, we provide two architectural design guidelines and corresponding solutions to ensure the functional correctness of SuperClocking in general SFQ devices.By applying our clocking scheme, synchronizer, and guidelines to the latest general-purpose SFQ CPU, SuperSFQ achieves up to 62.5 times higher frequency and improves single-thread and multithread performance by 17 and 62.5 times, respectively, compared to conventional designs, with only 34.4% Josephson junction overhead.In addition, to demonstrate the generality of SuperSFQ, we apply SuperSFQ to 48 different benchmark circuits, achieving 88.5 times higher frequency compared to conventional designs, on average.
Rapid single flux quantum (RSFQ) logic is a digital circuit technology that in recent years has presented itself as an alternative to semiconductors in the application of ultra high speed, very low power applications. The optimal timing of digital circuits operating at hundreds of GHz is still a complex problem for both RSFQ and semiconductor technologies. The fact that most RSFQ gates require a clock signal to function makes this even more complex. Various RSFQ timing schemes have been adapted from semiconductor design methodologies, and some have been designed specifically for RSFQ. Currently, synchronous clocking schemes outperform other schemes, but with the scale of RSFQ circuits ever increasing, the proper use of timing schemes are becoming more crucial. This paper describes a new asynchronous self-timing scheme where the details of clock distribution and clocking are built into the logic gates. Tests were done on the newly developed asynchronous logic gates and an asynchronous full adder was implemented and tested
We describe the development of single-flux-quantum (SFQ) microprocessors and the related technologies such as designing, circuit architecture, microarchitecture, etc. Since the microprocessors studied here aim for a general-purpose computing system, we employ the complexity-reduced (CORE) architecture in which the high-speed nature of the SFQ circuits is used not for increasing processor performance but for reducing the circuit complexity. The bit-serial processing is the most suitable way to realize the CORE architecture. We assembled all the best technologies concerning SFQ integrated circuits and designed the SFQ microprocessors, CORE1α, CORE1β, and CORE1γ. The CORE1β was made up of about 11000 Josephson junctions and successfully demonstrated. The peak performance reached 1400 million operations per second with a power consumption of 3.4mW. We showed that the SFQ microprocessors had an advantage in a performance density to semiconductor's ones, which lead to the potential for constructing a high performance SFQ-circuit-based computing system.
An architecture of rapid single-flux-quantum logic circuits using clockless gates is proposed. Compared with a circuit composed of only clocked gates, a circuit using clockless gates includes fewer clocked gates, and therefore, a smaller clock distribution network. A circuit in the proposed architecture is constructed from pipeline stages. Each pipeline stage consists of a column of clocked gates and a succeeding block of clockless gates. The function of a pipeline register of a stage is implemented by the latching function of the clocked gates in the column. The combinational logic of a stage is implemented by the logic function of the clocked gates in the column and that of the clockless gates in the block. We designed a two-stage-pipelined 4-bit carry-lookahead-adder using clockless gates as a design example of the proposed architecture. It had reductions of the count of clocked gates and the area by 33 and 13%, respectively, compared with that composed of only clocked gates. The 4-bit adder in the proposed architecture was fabricated and demonstrated with the maximum clock frequency of 21.4 GHz and the bias margin of ±16%.
A novel method of design for controllers of pipelined microprocessors using single-flux-quantum (SFQ) logic has been proposed. The proposed design approach is based on one hot encoding and is very suitable for designing a finite state machine using SFQ logic circuits, where each internal state of the microprocessor is represented by a flip-flop. In this approach, decoding of the internal state can be performed instantaneously, in contrast to the case in the conventional method using a binary state register. Moreover, pipelining is effectively implemented without increasing the circuit size because no pipeline registers are required in the one hot encoding. By using this method, we have designed a controller for our new SFQ microprocessors, which employs pipelining. The number of Josephson junctions of the newly designed controller is 1067, while the previous version without pipelining contains 1721 Josephson junctions. These results indicate that the proposed design approach is very effective for pipelined SFQ microprocessors. We have implemented a new controller using the NEC 2.5 kA cm −2 Nb standard process and confirmed its correct operation experimentally.
We propose an asynchronous single flux quantum (SFQ) logic gate composed of a conventional clocked logic gate and an input signal merger that generates a self-clock for the clocked logic gate. By using this self-clocking, we can design any 0-preserving logic gate, including exclusive OR (XOR). The proposed asynchronous SFQ logic gates have adjustable allowable input skew by varying the delay between the merger and internal clock input for the clocked logic gate. We design and evaluate the asynchronous AND, OR, and XOR gates that have an allowable input skew of approximately 8 ps assuming the use of the 10 kA/cm2Nb fabrication process using the proposed design method. We experimentally confirm the correct operation of the asynchronous SFQ XOR gate. We design a completely asynchronous 4-bit-input carry-lookahead adder (CLA) using the designed asynchronous gates. We found that the circuit area and number of Josephson junctions required to design a 4-bit-input CLA can be reduced by > 50% compared to that of the synchronous concurrent-flow 4-bit-input CLA without deterioration of the throughput.