Approximate computing is an emerging paradigm to improve the energy efficiency for error-tolerant applications. Many iterative approximate logic synthesis (ALS) methods were proposed to automatically design approximate circuits. However, as the sizes of circuits grow, the runtime of ALS grows rapidly. Thus, a crucial challenge is to ensure circuit quality while improving the efficiency of ALS. This work proposes a dual-phase iterative framework to accelerate the iterative ALS flows. In the first phase, a comprehensive circuit analysis is performed to gather the necessary information, including the error information. In the second phase, minimal incremental computation is employed based on the information from the first phase. The experimental results show that the proposed method achieves an acceleration by up to 21.8 × without loss of circuit quality compared to the state-of-the-art methods.

Approximate computing emerges as an energy-efficient computing paradigm designed for applications that can tolerate errors. Many iterative methods for approximate logic synthesis (ALS) have been developed to automatically synthesize approximate circuits. Nonetheless, most of them overlook the potential of applying multiple local approximate changes (LACs) simultaneously in one iteration, which can significantly reduce the overall computation time. In this article, we propose AccALS 2.0, a novel framework for further accelerating iterative ALS flows, which is based on simultaneous selection of multiple LACs in a single round. However, there are two challenges for selecting multiple LACs. The first is that the mutual influence of multiple LACs can affect the estimation of the circuit error. The second is that there may exist conflicts among multiple LACs. To address these issues, first, we propose an efficient measure for the mutual influence between two LACs. With its help, we transform the problems of solving the LAC conflicts and selecting multiple LACs into a unified maximum independent set problem for solving. The experimental results showed that AccALS 2.0 outperforms state-of-the-art ALS methods in runtime, while achieving similar or better-circuit quality.

Power consumption has become the major concern in neural network accelerators for edge devices. The novel non-volatile-memory (NVM) based computing-in-memory (CIM) architecture has shown great potential for better energy efficiency. However, most of the recent NVM-CIM solutions mainly focus on fixed-point calculation and are not applicable to floating-point (FP) processing. In this paper, we propose an analog-domain floating-point CIM architecture (AFPR-CIM) based on resistive random-access memory (RRAM). A novel adaptive dynamic-range FP-ADC is designed to convert the analog computation results into FP codes. Output current with high dynamic range is converted to a normalized voltage range for readout, to prevent precision loss at low power consumption. Moreover, a novel FP-DAC is also implemented which reconstructs FP digital codes into analog values to perform analog computation. The proposed AFPR-CIM architecture enables neural network acceleration with FP8 (E2M5) activation for better accuracy and energy efficiency. Evaluation results show that AFPR-CIM can achieve 19.89 TFLOPS/W energy efficiency and 1474.56 GOPS throughput. Compared to traditional FP8 accelerator, digital FP-CIM, and analog INT8-CIM, this work achieves 4.135×, 5.376×, and 2.841× energy efficiency enhancement, respectively.

Storing pre-computed results of frequently-used functions into lookup table (LUT) is a popular way to improve energy efficiency, but its advantage diminishes as the number of input bits increases. A recent work shows that by decomposing the target function approximately, the total LUT entries can be dramatically reduced, leading to significant energy saving. However, its heuristic approximate decomposition algorithm leads to sub-optimal approximation quality. Also, its rigid hardware architecture only supports disjoint decomposition and may have unnecessary extra power consumption sometimes. To address these issues, we develop a novel approximate decomposition algorithm based on beam search and simulated annealing, which can reduce 11.1% approximation error. We also propose a non-disjoint approximate decomposition method and two reconfigurable architectures. The first has 10.4% less error using 19.2% less energy and the second has 23.0% less error with same energy consumption compared to the state-of-the-art design.

Approximate computing is an emerging computing paradigm to design energy-efficient systems. Many greedy approximate logic synthesis (ALS) methods have been proposed to automatically synthesize approximate circuits. They typically need to consider all local approximate changes (LACs) in each iteration of the ALS flow to select the best one, which is time-consuming. In this paper, we propose SEALS, a Sensitivity-driven Efficient ALS method to speed up a greedy ALS flow. SEALS centers around a newly proposed concept called sensitivity, which enables a fast and accurate error estimation method and an efficient method to filter out unpromising LACs. SEALS can handle any statistical error metric. The experimental results show that it outperforms a state-of-the-art ALS method in runtime by 12X to 15X without reducing circuit quality.

With the slow-down of Moore's law and the increasing requirements on energy efficiency, alternative logic styles compared to complementary static CMOS have to be revisited for digital circuit implementations. Pass Transistor Logic (PTL) gained much attention in the '90s, however, only a limited number of recent investigations and publications regarding PTL exist that use advanced technology nodes. This paper compares key performance metrics of 22 different PTL based 1-bit full adder designs to a complementary static CMOS logic reference, using a recent 12nm FinFET technology. The figures of merit are the propagation delay, the energy consumption, and the energy-delay-product (EDP). Our investigations show that PTL based adder circuits can have an up to 49% decreased delay and a 48% and 63% reduced energy consumption and EDP, respectively, compared to a state-of-the-art complementary CMOS logic reference. In addition, we analyzed the impact of PVT variations on the delay for selected PTL full adder designs.

Recent advanced high-speed communication systems, such as optical systems, require highest reliability at lowest possible power consumption. Thus, Pass Transistor Logic (PTL) is gaining lots of interest in these communication systems due to its power saving potential compared to traditional CMOS logic. However, due to the non-conventional logic structure, its susceptibility to radiation-induced soft errors is different from CMOS circuitry. Due to the unique generation and propagation of Single Event Transients (SETs) in PTL, different approaches for PTL soft error rate (SER) estimation are required. In this paper we propose a machine learning (ML) approach for SET propagation in PTL logic. Multi-layer feed-forward neural network together with support vector classifier (SVC) are used to build the SET pulse width and pulse amplitude models. Bayesian optimization using Gaussian Processes is utilized to tune the hyperparameters of neural network. The experimental results on full adder (FA), which is the key component in many large cirucits such as ALU, and comparison with Monte Carlo (MC) spectre simulations confirm the accuracy and speed of the proposed method.

Approximate computing is an emerging strategy to improve the energy efficiency of many error-tolerant applications. To design an approximate circuit automatically, many approximate logic synthesis (ALS) methods have been proposed, among which many are greedy. To improve the synthesis quality of these greedy methods, one key is to calculate the errors of all candidate approximate transformations accurately. However, the traditional simulation-based method is time consuming. Instead, many existing methods just perform quick but inaccurate error estimation. In this work, to improve both the accuracy and runtime of error estimation, we propose VECBEE, a versatile efficiency–accuracy configurable batch error estimation method for greedy ALS. It is based on Monte Carlo simulation and an efficient technique to capture whether a signal change due to an introduced approximation will be propagated to each primary output. VECBEE is generally applicable to any statistical error measurement, such as error rate and average error magnitude, and any graph-based circuit representation. It allows a flexible tradeoff between the error estimation accuracy and the runtime, while even the fully accurate version is much faster than the traditional simulation-based method. We apply VECBEE to two representative greedy ALS methods and demonstrate its effectiveness in generating better approximate circuits. The code of VECBEE is made open source.

Convolutional neural network (CNN) based machine learning requires a highly parallel as well as low power consumption (including leakage power) hardware accelerator. In this paper, we will present a digital ReRAM crossbar based CNN accelerator that can achieve significantly higher throughput and lower power consumption than state-of-arts. The CNN is trained with binary constraints on both weights and activations such that all operations become bitwise. With further use of 1-bit comparator, the bitwise CNN model can be naturally realized on a digital ReRAM-crossbar device. A novel sneak-path-free ReRAM-crossbar is further utilized for large-scale realization. Simulation experiments show that the bitwise CNN accelerator on the digital ReRAM crossbar achieves 98.3% and 91.4% accuracy on MNIST and CIFAR-10 benchmarks, respectively. Moreover, it has a peak throughput of 792GOPS at the power consumption of 6.3mW, which is 18.86 times higher throughput and 44.1 times lower power than CMOS CNN (non-binary) accelerators.

The recently emerging resistive random-access memory (RRAM) can provide nonvolatile memory storage but also intrinsic computing for matrix-vector multiplication, which is ideal for the low-power and high-throughput data analytics accelerator performed in memory. However, the existing RRAM crossbar--based computing is mainly assumed as a multilevel analog computing, whose result is sensitive to process nonuniformity as well as additional overhead from AD-conversion and I/O. In this article, we explore the matrix-vector multiplication accelerator on a binary RRAM crossbar with adaptive 1-bit-comparator--based parallel conversion. Moreover, a distributed in-memory computing architecture is also developed with the according control protocol. Both memory array and logic accelerator are implemented on the binary RRAM crossbar, where the logic-memory pair can be distributed with the control bus protocol. Experimental results have shown that compared to the analog RRAM crossbar, the proposed binary RRAM crossbar can achieve significant area savings with better calculation accuracy. Moreover, significant speedup can be achieved for matrix-vector multiplication in neural network--based machine learning such that the overall training and testing time can be both reduced. In addition, large energy savings can be also achieved when compared to the traditional CMOS-based out-of-memory computing architecture.

Emerging resistive random-access memory (RRAM) can provide non-volatile memory storage but also intrinsic logic for matrix-vector multiplication, which is ideal for low-power and high-throughput data analytics accelerator performed in memory. However, the existing RRAM-based computing device is mainly assumed on a multi-level analog computing, whose result is sensitive to process non-uniformity as well as additional AD- conversion and I/O overhead. This paper explores the data analytics accelerator on binary RRAM-crossbar. Accordingly, one distributed in-memory computing architecture is proposed with design of according component and control protocol. Both memory array and logic accelerator can be implemented by RRAM-crossbar purely in binary, where logic-memory pairs can be distributed with protocol of control bus. Based on numerical results for fingerprint matching that is mapped on the proposed RRAM-crossbar, the proposed architecture has shown 2.86x faster speed, 154x better energy efficiency, and 100x smaller area when compared to the same design by CMOS-based ASIC.

On-line machine learning has become the need for future data analytics. This work will show an ℓ2norm based hardware solver for on-line machine learning that can significantly reduce training time when compared to the traditional gradient-based solution using backward propagation. We will show that the intensive matrix-vector multiplication in ℓ2norm solution can be mapped onto a distributed in-memory accelerator using the recent resistive switching random access memory (RRAM) device. A digitized matrix-vector multiplication accelerator will be developed based on the distributed RRAM-crossbar. Such a distributed RRAM-crossbar architecture can utilize the reformulated ℓ2norm solver with a scalable and energy-efficient solution for real-time training and testing in image recognition. Experiment results have shown that significant speedup can be achieved for matrix-vector multiplication in the ℓ2norm solver such hat the overall training and testing time can be reduced respectively. In addition, large energy saving can be also achieved when compared to the traditional CMOS-based out-of-memory computing architecture.

It is challenging to efficiently evaluate the performance bound of high-precision analog circuits with input and parameter variations at nano-scale. With the use of zonotope to model uncertainty of input data pattern (or jitter) and multiple parameters, a reachability-based verification is developed in this paper to compute the worst-case eye-diagram. The proposed zonotope-based reachability analysis can consider both spatial and temporal variations in one-time simulation. Moreover, a nonlinear zonotoped macromodeling is further developed to reduce the computational complexity. Performance bound for I/O links considering the parameter variations are evaluated. In addition, the eye-diagrams are generated by the proposed zonotoped macromodel for performance evaluation considering both temporal and spatial variations. As shown by experiments, the zonotoped macromodel achieves up to 450× speedup compared to the Monte Carlo simulation of the original model within small error under specified macromodel order for high-speed I/O links eye-diagram verification.

Big-data storage poses significant challenges to anonymization of sensitive information against data sniffing. Not only will the encryption bandwidth be limited by the I/O traffic, the transfer of data between the processor and the memory will also expose the input-output mapping of intermediate computations on I/O channels that are susceptible to semi-invasive and non-invasive attacks. Limited by the simplistic cell-level logic, existing logic-in-memory computing architectures are incapable of performing the complete encryption process within the memory at reasonable throughput and energy efficiency. In this paper, a block-level in-memory architecture for advanced encryption standard (AES) is proposed. The proposed technique, called DW-AES, maps all AES operations directly to the domain-wall nanowires. The entire encryption process can be completed within a homogeneous, high-density, and standby-power-free non-volatile spintronic-based memory array without exposing the intermediate results to external I/O interface. Domain-wall nanowire-based pipelining and multi-issue pipelining methods are also proposed to increase the throughput of the baseline DW-AES with an insignificant area overhead and negligible difference on leakage power and energy consumption. The experimental results show that DW-AES can reduce the leakage power and area by the orders of magnitude compared with existing CMOS ASIC accelerators. It has an energy efficiency of 22 pJ/b, which is 5× and 3× better than the CMOS ASIC and memristive CMOL-based implementations, respectively. Under the same area budget, the proposed DW-AES achieves 4.6× higher throughput than the latest CMOS ASIC AES with similar power consumption. The throughput improvement increases to 11× for pipelined DW-AES at the expense of doubling the power consumption.

The emerging resistive random-access-memory (RRAM) crossbar provides an intrinsic fabric for matrix-vector multiplication, which can be leveraged as power efficient linear embedding hardware for data analytics such as compressive sensing. As the matrix elements are represented by resistance of RRAM cells, it imposes constraints for the embedding matrix due to limited RRAM programming resolution. A random Boolean embedding can be efficiently mapped to the RRAM crossbar but suffers from poor performance. Learning-based embedding matrices can deliver optimized performance but are continuous-valued which prevents it from being mapped to RRAM crossbar structure directly. In this paper, we have proposed one algorithm that can find an optimal Boolean embedding matrix for a given learned real-valued embedding matrix, so that it can be effectively mapped to the RRAM crossbar structure while high performance is preserved. The numerical experiments demonstrate that the proposed optimized Boolean embedding can reduce the embedding distortion by 2.7x, and image recovery error by 2.5x compared to the random Boolean embedding, both mapped on RRAM crossbar. In addition, optimized Boolean embedding on RRAM crossbar exhibits 10x faster speed, 17x better energy efficiency, and three orders of magnitude smaller area with slight accuracy penalty, when compared to the optimized real-valued embedding on CMOS ASIC platform.