Bloom filters can be used in network intrusion detection systems to detect known attack signatures in packet payloads. In this paper we propose and analyze the potential application of superconductor flux quantum technology for streaming data inspection with Bloom filters designed with Reciprocal Quantum Logic (RQL). This paper describes the gate-level design, performance, and energy-efficiency analysis of three superconductor 2 Kbit Bloom filters with 1) the run-time selection of the number of hashes per stream, and 2) different numbers of input streams per Bloom filter. The Bloom filter circuits were designed using a bottom-up approach with manual placing and routing of basic RQL gates. The design complexity is below 97K Josephson junctions. The highest clock frequency reached in the simulation of the circuits is 14.7 GHz. The false positive rates of the RQL Bloom filters are in very close agreement with the theoretical expectations of the false positive probability for the filters. For the cryocooling efficiency of 0.1 percent, the RQL Bloom filters demonstrate high energy efficiency in the range of ~1.5-43.6 pJ/stream/operation at room temperature for stream lengths from 16 to 256 bits. All circuits are designed and simulated for the 248 nm MIT Lincoln Laboratory SFQ5ee fabrication process.
Bloom filters can be used in network intrusion detection systems to detect known attack signatures in packet payloads. In this paper we propose and analyze the potential application of superconductor flux quantum technology for streaming data inspection with Bloom filters designed with Reciprocal Quantum Logic (RQL). This paper describes the gate-level design, performance, and energy-efficiency analysis of three superconductor 2 Kbit Bloom filters with 1) the run-time selection of the number of hashes per stream, and 2) different numbers of input streams per Bloom filter. The Bloom filter circuits were designed using a bottom-up approach with manual placing and routing of basic RQL gates. The design complexity is below 97K Josephson junctions. The highest clock frequency reached in the simulation of the circuits is 14.7 GHz. The false positive rates of the RQL Bloom filters are in very close agreement with the theoretical expectations of the false positive probability for the filters. For the cryocooling efficiency of 0.1 percent, the RQL Bloom filters demonstrate high energy efficiency in the range of ~1.5-43.6 pJ/stream/operation at room temperature for stream lengths from 16 to 256 bits. All circuits are designed and simulated for the 248 nm MIT Lincoln Laboratory SFQ5ee fabrication process.
Superconductor single flux quantum (SFQ) technology is one of the promising candidates for energy-efficient high-performance computing. A new generation of SFQ logics, Reciprocal Quantum Logic (RQL) with no static power dissipation in bias resistors, offers an opportunity to dramatically decrease energy consumption in superconductor processors. Although several low complexity RQL processing units have already been demonstrated, the use of RQL for local storage design has not been explored yet. The objective of this dissertation is to design on-chip local storage structures such as memory, register files, and caches with RQL technology and analyze their energy efficiency, complexity, and performance characteristics. The physical chip design of these RQL storage units is not feasible at this point because both CAD tools for physical VLSI chip design and as well as a target fabrication process are under development at Northrop-Grumman Systems Corp. (Baltimore, MD) and MIT Lincoln Laboratory, respectively. In order to achieve our goal, the layout-aware cell-level design process using VHDL RQL cell library developed at the Ultra High Speed Computing Lab in Stony Brook University has been used. The SBU VHDL RQL cell library specifies the dynamic and stand-by energy consumption, latency, JJ complexity, and approximate sizes of individual cells based on the input received from the JJ-level RQL designers. Clock propagation skew and wire delays are accounted for during circuit simulation. The circuit simulation is done with Mentor Graphics design and verification tools. As a result of the work, key characteristics of the 8.5 GHz multiported RQL storage structures with their capacity in the range of 1-4 kbit have been determined. The average energy consumption of the RQL storage designs is ~3.0-9.5 fJ/bit/operation at room temperature and the cryo-cooling efficiency is 0.1%. The data from this dissertation also reveal the critical issues that need to be considered in the RQL storage design. These will be helpful for further development on superconductor VLSI design. | 124 pages
New superconductor single flux quantum (SFQ) technology, such as Reciprocal Quantum Logic (RQL), is currently considered one of the promising candidates for highperformance energy-efficient computing. This paper presents our work on the design and detailed energy efficiency analysis of three types of 32- and 64-bit RQL multi-ported pipelined local storage structures (13 total), namely 1) random access memory (RAM) and register files, 2) direct-mapped write-through and write-back caches, and 3) first-in-first-out (FIFO) buffers. Our layout-aware cell-level design process uses a VHDL RQL cell library developed at the Ultra High Speed Computing Laboratory at Stony Brook University (SBU). The SBU VHDL RQL cell library specifies the dynamic and standby energy consumption, gate delays, a number of Josephson junctions (JJs) per cell, and approximate sizes of individual cells based on the parameters of the 248 nm 100 μA/μm210 Nb metal layer SFQ fabrication process currently under development at the MIT Lincoln Laboratory. Gate and wire delays as well as clock skew are taken into account during digital circuit simulation done with Mentor Graphics CAD tools. After completing a physical chip layout, the circuit models need to be updated and re-simulated to include the effects of parasitic inductances and actual wire lengths on signal propagation delays. To meet both performance and energy efficiency targets, the RQL storage structures were designed with RQL non-destructive read-out single-bit storage cells. We chose a relatively moderate clock frequency of 8.5 GHz for all storage units to keep their read latencies in the range of 1- 3 cycles. The most complex design in terms of JJs is a tripleported 4 Kbit 64x64-bit register file with 253,918 JJs and its read access latency of 338 ps. The highest energy consumption in terms of energy/operation/bit (~9.5 aJ at 4.2 K) is for a write hit in a 2 Kbit 32-bit wide write-back cache. The average energy consumption of the RQL storage designs varies from ~1.6 aJ/operation/bit for a small 4x32-bit FIFO to 7.3 aJ/operation/bit for the 2 Kbit write-back cache at 4.2 K. Given the cryocooler efficiency of 0.1%, this means the energy consumption of ~1.6-7.3 fJ/operation/bit at room temperature. The physical implementation of the RQL storage units will become feasible upon the development of the target MIT fabrication process and CAD tools for VLSI RQL chip design in 2015-2016.
New superconductor single flux quantum logics with no static power dissipation in bias resistors, such as Reciprocal Quantum Logic (RQL), offer opportunities to create energy-efficient superconductor processors operating at high frequencies with ultra-low power consumption. This paper discusses the results of our work on the cell-level design and analysis of a benchmark set of 32-/64-bit RQL processor integer and floating-point units such as adders, multipliers, an arithmetic-logic unit, and an array shifter, as well as small 1-4 Kbit RQL on-chip storage components such as register files, on-chip memory, and the top level caches. Our layout-aware design process includes the complete cell-level design and approximate physical layout of the circuits followed by the VHDL simulation, verification, and energy profiling using our RQL VHDL cell library tuned to the future MIT Lincoln Laboratory 10 kA/cm2 248 nm process with 10 Nb metal layers and the minimum JJ critical current of 38 μA. Our designs have the energy efficiency of ~1.0 single-precision TFLOPS/W and '0.5 double-precision TFLOPS/W for floating-point units, and ~1-24 TOPS/W for 32-bit integer units at room temperature using the cryocooling efficiency of 0.1 % (1000 W/W). The 1-4 Kbit 32-/64-bit multi-ported scratchpad memory, register files, write-through and write-back caches designed with RQL Non-Destructive Read-Out storage cells have the average energy consumption of 3.0-9.5 fJ/bit/operation at room temperature using the cryocooling efficiency of 0.1%. While these results are very promising, more work is needed to evaluate the contribution of the energy costs of instruction scheduling and off-chip main memory access to the energy efficiency of RQL computing across a whole system.
Complimentary Medal-Oxide Semiconductor (CMOS) technology is expected to soon reach its fundamental limits of operation. The fundamental speed limit of about 4 GHz has already effectively been sidestepped by parallelization. This increases raw processing power but does nothing to improve power dissipation or latency. One approach for increasing computing performance involves using superconducting digital logic circuits. In this thesis I describe a new kind of superconducting logic, invented by Quentin Herr at Northrop Grumman, which uses reciprocal pairs of quantized single magnetic flux pulses to encode classical bits. In Reciprocal Quantum Logic (RQL) the data is encoded in integer units of the magnetic flux quantum. RQL gates operate without the bias resistors of previous superconducting logic families and dissipate several orders of magnitude less power. I demonstrate the basic operation of key RQL gates (AndOr, AnotB, Set/Reset) and show their self-resetting properties. Together, these gates form a universal logic set and provide memory capabilities. Experiments measuring the bit error rate of the AndOr gate extrapolated a minimum BER of 10−480 and a BER of 10−44 with 30% margins on flux biasing. I describe an analytic timing model for RQL gates which demonstrates the self-correcting timing features. From this model I derive equations for the timing behavior and operating limits. Using this timing model I ran simulations to determine correction factions for more accurate predictions at higher frequencies. Using these results, I also develop Very High Speed Integrated Circuit (VHSIC) Hardware Description Language (VHDL) models to describe the combinational logic of RQL gates. To test the timing predictions of the timing model, I performed three experiments on Nb/AlOx/Nb circuits at 4.2 K. The first measured the time of output. The second measured the operating margins of the circuit. The third measured the maximum frequency of operation for RQL circuits. Together, these three experiments showed quantitative agreement with the model for the timing output, qualitative agreement with the limits of operation, and a projected speed limit of 50 GHz for the Hypres 4.5 kA/cm2 process. To power RQL circuits I describe a new design for power splitters and combiners which minimize standing waves. I describe a new kind of Wilkinson power splitter which required numerical optimization but proved to be adequate. I experimentally tested two new designs of the power splitter. Both showed less than 10% variation in standing waves between power splitter and combiner, making it adequate for RQL circuits. I also compared these results with the S-parameters of the power network, which also indicated that the design was adequate for RQL circuits. Finally, I tested an 8-bit Kogge-Stone architecture carry-look ahead adder designed using VHDL models. The adder contained 815 Josephson junctions and was fully functional at 6.21 GHz with a latency of 1.25 clock cycles. The adder produced the correct logical output, had a measured optimal operating point within 8% of the optimal simulated operating point, and measured power margins of 1 dB. It operated best at the designed clock amplitude of 0.88 Ic and dissipated 0.570 mW of power.
Abstract : The major objective of the project was to design and demonstrate operation of key components of a 30 GHz 16-bit RSFQ processor prototype implemented with the AIST/ISTEC 10 kA/cm sq. fabrication process. Our team has developed complete logical and physical designs of five RSFQ chips using the CONNECT cell library and RSFQ CAD tools developed at the Universities of Yokohama and Nagoya (Japan). The major results are the world's first successful design, fabrication, and demonstration of correct operation of a 20 GHz 8x8-bit parallel carry-save RSFQ multiplier with approximately 6K JJs, a 16-bit sparse-tree wave-pipelined RSFQ adder with approximately 10K JJs, and partial operation of an 8-bit ALU chip with approximately 9K JJs. The goal of the second phase of the project was to get detailed understanding of the performance, complexity, and energy efficiency of on-chip storage units implemented with superconductor Reciprocal Quantum Logic (RQL) using our RQL VHDL cell library tuned to the MIT Lincoln Laboratory 10 kA/cm2 248 nm process. The 8.5 GHz 1-4 Kbit 32-/64-bit multi-ported scratchpad memory, register files, write-through and write-back caches designed with RQL Non-Destructive Read-Out storage cells have the average energy consumption of 3.0-9.5 fJ/bit/operation at room temperature using the cryocooling efficiency of 0.1%.
Josephson Junction (JJ) based superconductor logic families have been proposed and implemented to process analog and digital signals [1] for its low energy dissipation and ultrafast switching speed. Thanks to its construction of resistance-less wires and ultrafast switches, it can operate at clock frequencies of several tens of gigahertz and even hundreds of thousands of times as energy efficient as its CMOS counterparts. It has been perceived to be an important candidate to replace stateof-the-art CMOS due to the superior potential in operation speed and energy efficiency, as recognized by the U.S. IARPA C3 and SuperTools Programs and Japan MEXT-JSPS Project. The design and fabrication of superconducting circuits have already been established [2]-[4]. In addition, a prototype superconducting microprocessor "Core 1" has been demonstrated in 2004 [3], which is able to execute instructions at a high clock frequency of several tens of gigahertz, and with extremely low-power dissipation. These achievements make superconducting electronics highly promising for future high-performance computing applications. As one of the most matured superconducting technology, the Rapid-Single-Flux-Quantum (RSFQ) technology is proposed by K. Likharev, O. Mukhanoc, V. Semenov in 1985 [1]. Despite its capability to be operated at an ultra-high speed of hundreds of GHz while maintaining extremely low switching energy (10^-19 J), it suffers from an increasing static power due to on-chip resistors that are required for constant DC bias supply for the main RSFQ circuit. Numerous methods have been proposed to resolve the static power dissipation problem of RSFQ, including low-voltage RSFQ (LV-RSFQ) [5], reciprocal quantum logic (RQL) [6], LRbiased RSFQ [7] and energy-efficient single-flux quantum (eSFQ) [8]. The Adiabatic Quantum-Flux-Parametron (AQFP) technology, on the other hand, uses AC bias/excitation currents as both multiphase clock signal and power supply [9] to mitigate the power consumption overhead of DC bias while operating at a frequency of few GHz. Consequently, AQFP is remarkably energy efficient compared to RSFQ, albeit operating at a lower frequency. The energy-delay-product (EDP) of the AQFP circuits fabricated using processes such as the AIST standard process 2 (STP2) and the MIT-LL SFQ process [10], [11], is at least 200 times smaller than those of the other energy-efficient superconductor logics and is only three orders of magnitude larger than the quantum limit [9]. Physical testing results of an AQFP 8-bit carry-look-ahead adder and large scale circuits consisting up-to 10,000 AQFP logic gates have demonstrated the AQFP being a promising technology that is robust against circuit parameter variations [12]. Despite the high application potential of AQFP in VLSI circuits, a systematic, automatic synthesis framework for AQFP is imminent. There are two features of AQFP that restrict conventional CMOS synthesis methods being directly applied on AQFP. In spite of And-Or-Inverter(AOI) based representation, which conventional CMOS circuits highly relies on, AQFP circuits prefer majority gates. In fact, its two inputs AND and OR gates are also built with three inputs majority gate with one input being constant. In addition, given its clock-synchronized data propagation nature, AQFP technology requires all inputs to any gate having equal delay. In order to meet this balanced timing requirement, splitters and buffers need to be inserted to the circuit. As a matter of fact, some circuit size can be doubled even with optimum amount of buffers and splitters inserted. The buffer and splitter insertion method can have a huge impact on the overall resource consumption. As the design complexity increases, an unoptimized buffer and splitter insertion method could result in huge amount of unnecessary buffers and splitters added. In addition to a complete synthesis framework, an Integrated Development Environment (IDE) for AQFP design is also lacking. It is imminent to have an IDE for AQFP integrating tools that offer schematic and layout editor, simulation and analysis for better and more efficient AQFP design flow. In this paper, we propose a complete design tool for AQFP design including an Integrated Development Environment (IDE), a complete majority based synthesis framework and a buffer and splitter insertion framework. we propose a majority gates synthesis framework for AQFP circuits that is capable of converting any AOI netlist to its corresponding MAJ netlist by mapping all feasible three-input sub-netlists to corresponding MAJ based implementations. In addition, we also propose an automated buffer and splitter insertion method that is capable of adding the optimum amount of buffers and splitters to any given gate-level netlist. The proposed method can find the minimum amount of buffers and splitters to inserted to achieve equal delay with any library limitation on the size of splitters. Experimental results suggest that the proposed methods can deliver very optimized results. The majority conversion tool can reduce circuit size by an average of 16:47% and delay by 30:21% in average. The buffer and splitter insertion tool only introduces an average of 14:24% overhead in size and 4:70% delay with splitter fanout size limited to 4 compared to an unachievable ideal results with no limitations on splitter fan-out size. Overall, the balanced design in majority gates can be reduced by 23:85% and 29:54% in size and delay compared to its AND/OR/Inverter implemented counterpart.