Resistive random access memory (RRAM)-based hardware accelerators are playing an important role in the implementation of in-memory computing (IMC) systems for artificial intelligence applications. The latter heavily rely on vector-matrix multiplication (VMM) operations that can be efficiently boosted by RRAM devices. However, the stochastic nature of the RRAM technology is still challenging real hardware implementations. To study the accuracy degradation of consecutive VMM operations, in this work we programed two RRAM subarrays composed of $8\times $ 8 one-transistor-one-resistor (1T1R) cells following two different distributions of conductive levels. We analyze their robustness against 1000 identical consecutive VMM operations and monitor the inherent devices’ nonidealities along the test. We finally quantize the accuracy loss of the operations in the digital domain and consider the trade-offs between linearly distributing the resistive states of the RRAM cells and their robustness against nonidealities for future implementation of IMC hardware systems.

Convolutional Neural Networks (CNNs) have revolutionized many applications in recent years, especially in image classification, video processing, and pattern recognition. This success of CNNs has been a motivating factor for solving even more complex problems involving multiple data modalities. Traditionally, a single CNN accelerator has been optimized for just one task or has been used to perform correlated tasks. We leverage the CNNs capability to learn patterns and use one accelerator to perform multiple uncorrelated tasks from different modalities and achieve an average accuracy above 90%, which would otherwise require three accelerators. Two types of CNN architectures (i.e., fused and branched) are evaluated for three distinct tasks based on accuracy, quantization, pruning, hardware resource utilization, power, and latency. Capitalizing on this, we have further proposed a runtime reconfigurable CNN accelerator supporting fault-tolerant (FT), high-performance (HP), and de-stress (DS) modes.

Simulation-based fault injection is a widely used technique for early-stage circuit reliability analysis. However, it consumes significant time, particularly for complex circuits. This paper introduces two Machine Learning (ML) methodologies to predict simulation-based fault injection outcomes at the gate level. The initial approach employs Neural Networks (NNs), extracting structural features from synthesis reports and simulation-related characteristics from Value Change Dump (VCD) waveforms. Nevertheless, NNs are restricted to learning from individual gate attributes. To exploit the comprehensive structure of entire circuits, we propose a method to convert circuits into graphs. This facilitates the utilization of Graph Neural Networks (GNNs) as advanced models, resulting in improved prediction performance. We select six open-source circuits with diverse complexities and functions to validate these methodologies and explore their adaptability across various circuits. Our experiments demonstrate the superior performance of GNNs compared to NNs in terms of prediction accuracy, efficiency in hyperparameter search, and the ability to address imbalanced datasets. Additionally, we investigate the feasibility of deploying the trained models to predict results in new circuits. Based on the experimental outcomes, we present an approach for leveraging the proposed methodology to accelerate simulation-based fault injection.

Nanometer circuits are becoming increasingly susceptible to soft errors. Selective hardening is a less expensive technique to improve the reliability of circuits because it hardens the critical components instead of hardening an entire circuit. One challenge of selective hardening is efficiently and effectively identifying the critical parts in circuits. Simulation-based fault injection is commonly used but extremely time-consuming, especially for complex circuits. This paper proposes an approach based on Graph Neural Networks (GNNs) to identify critical flip-flops in circuits. GNNs can take advantage of the circuit’s structural features and the features of individual flip-flops. To convert the features into abstract data that can be fed into GNNs, we provide a feature extraction method that uses a graph model to represent the relevant features of the circuit. The method converts the target circuit into a graph representing its architecture. The graph also contains features of individual flip-flops extracted from the circuit’s netlist and the Value Change Dump (VCD) waveforms of the test used for fault simulation. Additionally, we extract edge features in the graph to utilize the information on combinational gates on the path between flip-flops. Datasets generated based on two open-source RISC-V cores are used to validate the proposed approach. We compare the performance of different GNNs on them and discuss the contribution of edge features to their performance. Our experiments show that the prediction accuracy increases significantly with edge features. GraphSAGE and SAGE-GCN with edge features perform best among the selected GNNs. The highest accuracy we achieved on Ibex and RI5CY is 97.75% and 98.67%, respectively. We also provide a method to accelerate the process of critical flip-flop identification.

Asignificant rise in Artificial Intelligence (AI) has impacted many applications around us, so much so that AI has now been increasingly used in safety-critical applications. AI at the edge is the reality, which means performing the data computation closer to the source of the data, as opposed to performing it on the cloud. Safety-critical applications have strict reliability requirements; therefore, it is essential that AI models running on the edge (i.e., hardware) must fulfill the required safety standards. In the vast field of AI, Deep Neural Networks (DNNs) are the focal point of this survey as it has continued to produce extraordinary outcomes in various applications i.e. medical, automotive, aerospace, defense, etc. Traditional reliability techniques for DNNs implementation are not always practical, as they fail to exploit the unique characteristics of the DNNs. Furthermore, it is also essential to understand the targeted edge hardware because the impact of the faults can be different in ASICs and FPGAs. Therefore, in this survey, first, we have examined the impact of the fault in ASICs and FPGAs, and then we seek to provide a glimpse of the recent progress made towards the fault-tolerant DNNs. We have discussed several factors that can impact the reliability of the DNNs. Further, we have extended this discussion to shed light on many state-of-the-art fault mitigation techniques for DNNs.

The ImageNet moment was a turning point for Convolutional Neural Networks (CNNs), as it demonstrated their potential to revolutionize computer vision tasks. This triumph of CNNs has motivated solving even more complex problems involving multiple tasks from multiple data modalities. Conventionally, a single CNN accelerator has been optimized to perform just one task or multiple correlated tasks. This study presents a shared-layers approach that leverages the pattern-learning capabilities of CNNs to perform multiple uncorrelated tasks from different modalities using a single hardware accelerator. We overcame the challenge of data imbalance in multi-modal learning by synthetic data generation. We achieved an average classification accuracy above 90% on a single CNN accelerator, which would otherwise require three accelerators. Due to the reliability concerns imposed by transistor shrinking and aging, we extended the shared layers methodology and introduced a fault-tolerant CNN accelerator with reconfigurable capabilities supporting fault-tolerant (FT), high-performance (HP), and de-stress (DS) modes. FT mode provides high reliability against soft errors utilizing double/triple modular redundancy, HP mode offers peak performance of 0.979 TOPs using parallel execution, and DS mode reduces dynamic power consumption by up to 68.6% in clock-gated design and even more using a partial reconfiguration method, contributing to decelerating the aging process of the circuit. We have comprehensively evaluated two different CNN architectures (i.e., fused and branched), for three distinct tasks, in three different operating modes, based on accuracy, quantization, pruning, hardware resource utilization, power, energy, performance, and reliability.