Conventional heterogeneous computing systems built on PCIe interconnects suffer from inefficient fine-grained host-device interactions and complex programming models. In recent years, many proprietary and open cache-coherent interconnect standards have emerged, among which compute express link (CXL) prevails in the open-standard domain after acquiring several competing solutions. Although CXL-based coherent heterogeneous computing holds the potential to fundamentally transform the collaborative computing mode of CPUs and XPUs, research in this direction remains hampered by the scarcity of available CXL-supported platforms, immature software/hardware ecosystems, and unclear application prospects. This paper presents Cohet, the first CXL-driven coherent heterogeneous computing framework. Cohet decouples the compute and memory resources to form unbiased CPU and XPU pools which share a single unified and coherent memory pool. It exposes a standard malloc/mmap interface to both CPU and XPU compute threads, which share a single per-process page table for user applications, leaving the OS dealing with smart memory allocation, page auto-migration, and management of heterogeneous resources. This design significantly simplifies heterogeneous parallel programming to a level comparable to homogeneous programming. To facilitate Cohet research, we also present a fullsystem cycle-level simulator named SimCXL, which is capable of modeling all CXL sub-protocols and device types. SimCXL has been rigorously calibrated against a real CXL testbed with various CXL memory and accelerators, showing an average simulation error of 3 %. Our evaluation reveals that CXL.cache reduces latency by 68 % and increases bandwidth by$14.4 \times$compared to DMA transfers at cacheline granularity. Building upon these insights, we demonstrate the benefits of Cohet with two killer apps, which are remote atomic operation (RAO) and remote procedure call (RPC). Compared to PCIe-NIC design, CXL-NIC achieves a 5.5 to$40.2 \times$speedup for RAO offloading and an average speedup of$\mathbf{1. 8 6} \times$for$\mathbf{R P C}$(de)serialization offloading.

Compute eXpress Link (CXL) has emerged as a key enabler of memory disaggregation for future heterogeneous computing systems to expand memory on-demand and improve resource utilization. However, CXL is still in its infancy stage and lacks commodity products on the market, thus necessitating a reliable system-level simulation tool for research and development. In this paper, we propose CXL-DMSim1, an open-source full-system simulator to simulate CXL disaggregated memory systems with high fidelity at a gem5-comparable simulation speed. CXL-DMSim incorporates a flexible CXL memory expander model along with its associated device driver, and CXL protocol support with CXL.io and CXL.mem. It can operate in both app-managed mode and kernel-managed mode, with the latter using a dedicated NUMA-compatible mechanism. The simulator has been rigorously verified against a real hardware testbed with both FPGA- and ASIC-based CXL memory devices, which demonstrates the qualification of CXL-DMSim in simulating the characteristics of various CXL memory devices at an average simulation error of 3.4%. The experimental results using LMbench and STREAM benchmarks suggest that the CXL-FPGA memory exhibits a ~2.88× higher latency than local DDR while the CXL-ASIC latency is ~2.18×; CXL-FPGA achieves 45-69% of local DDR memory bandwidth, whereas the number for CXL-ASIC is 82-83%. The study also reveals that CXL memory can significantly enhance the performance of memory-intensive applications, improved by 23× at most with limited local memory for Viper key–value database and approximately 60% in memory-bandwidth-sensitive scenarios such as MERCI. Moreover, the simulator’s observability and expandability are showcased with detailed case-studies, highlighting its great potential for research on future CXL-interconnected hybrid memory pool.

The rapid growth of model sizes in advanced artificial intelligence algorithms, particularly in Transformerbased large language models (LLMs), has led to significant computational overhead. Mixture-of-Expert (MoE) models offer a solution through their sparsely gating mechanism but introduce new challenges of extensive all-to-all communication and model computational inefficiencies. This paper presents Hydra, a software/hardware co-design aimed at accelerating MoE inference on chiplet-based architectures. In software, Hydra employs a popularity-aware expert mapping strategy to optimize interchiplet communication. In hardware, it incorporates Content Addressable Memory (CAM) to eliminate expensive explicit token (un)-permutation based on sparse matrix multiplications and a redundant-calculation-skipping softmax engine to bypass unnecessary division and exponential operations. Evaluated in 22 nm technology, Hydra achieves latency reductions of $14.2 \times$ and $3.5 \times$ and power reductions of $169.1 \times$ and $18.9 \times$ over GPU and state-of-the-art MoE accelerator, respectively, thereby offering a scalable and efficient solution for MoE model deployment.

The victim cache was originally designed as a secondary cache to handle misses in the L1 data (L1D) cache in CPUs. However, this design is often sub-optimal for GPUs. Accessing the high-latency L1D cache and its victim cache can lead to significant latency overhead, severely degrading the performance of certain applications. We introduce HIVE, a high-priority victim cache designed to accelerate GPU memory accesses. HIVE handles memory requests first, before they reach the L1D cache. Our experimental results show that HIVE achieves an average performance improvement of $\mathbf{7 7. 1 \%}$ and $\mathbf{2 1. 7 \%}$ compared to the baseline and the state-of-the-art architecture, respectively.

Mixture-of-experts (MoE) technique holds significant promise for scaling up Transformer models. However, the data transfer overhead and imbalanced workload hinder efficient deployment. This work presents PIMoE, a heterogeneous system combining processing-in-memory (PIM) and neural-processing-unit (NPU) to facilitate efficient MoE Transformer inference. We propose a throttle-aware task offloading method that addresses workload imbalance between NPU and PIM, achieving optimal task distribution. Furthermore, we design a near-memory-controller data condenser to address the mismatch of sparse data layout between NPU and PIM, enhancing data transfer efficiency. Experimental results demonstrate that PIMoE achieves $4.5 \times$ speedup and $13.7 \times$ greater energy efficiency compared to the A 100, and $1.4 \times$ speedup over a state-of-the-art MoE platform.

In recent years, new methods for solving partial differential equations (PDEs) such as Monte Carlo random walk methods have gained considerable attention. However, due to the lack of hardware-intrinsic randomness in the conventional von Neumann architecture, the performance of PDE solvers is limited. In this paper, we introduce NeuroPDE, a hardware design for neuromorphic PDE solvers that utilizes emerging spintronic and ferroelectric devices. NeuroPDE incorporates spin neurons that are capable of probabilistic transmission to emulate random walks, along with ferroelectric synapses that store continuous weights non-volatilely. The proposed NeuroPDE achieves a squared error of less than 1e-2 compared to analytical solutions when solving diffus3.48× to 315× speedup in execution time and an energy consumption advantage of 2.7× to 29.8× over advanced CMOS-based neuromorphic chips. By leveraging the inherent physical stochasticity of emerging devices, this study paves the way for future probabilistic neuromorphic computing systems.

3D Gaussian Splatting (3DGS)–based rendering has gained increasing attention due to its state-of-the-art quality and broad applications in Augmented and Virtual Reality (AR/VR). However, the deployment of 3DGS in edge systems (~10FPS) faces challenge in achieving real-time (≥30FPS) performance due to computational resource constraints. Profiling reveals that the Gauss-Tile rasterization critically impacts pipeline efficiency due to its decoupled nature and computation-accuracy trade-off, which current hardware devices cannot effectively resolve. To this end, we propose a software-hardware co-design that performs rasterization via pattern matching. Our solution adopts flood encoding to represent tile coverage efficiently. It integrates a pattern-aware rasterization unit (PRU) and compresses the codebook using k-means clustering, preserving accuracy with minimal area overhead. At the architectural level, GauPRE adopts early depth-sorting and tile-group-wise rasterization to fuse coverage testing and alpha blending, enabling seamless data flow. We further integrated GauPRE with the GPU to support the end-to-end 3DGS rendering pipeline. Results demonstrate that the GPU+GauPRE delivers an 11.6× end-to-end speedup over the Jetson Orin Nano with only 0.1% area overhead. Meanwhile, the standalone GauPRE rasterization engine achieves 2.09× higher throughput compared to a state-of-the-art 3DGS accelerator.

3D-stacked DRAM is a key technology enabling the development of large language models (LLMs). However, the intensive computational demands lead to substantial energy consumption resulting from large-scale data movement. Integrating processing-in-memory (PIM) within DRAM has been employed to reduce data movement, but it elevates manufacturing costs and incurs additional area and power consumption. On the other hand, implementing processing-near-memory (PNM) outside the DRAM controller fails to eliminate the high energy consumption associated with interconnects during data transfer. To address these challenges, we propose a processing-in-controller (PIC) architecture aimed at 3D-stacked DRAM for efficient data movement. A size-scalable General Matrix-Vector Multiplication (GEMV) structure is proposed, supporting configurations ranging from 16 × 16 to 128 × 128, thereby enhancing its adaptability for a variety of applications. To address the low throughput bottleneck caused by DDR read latency, a ping-pong buffer with weight prefetching is introduced in the PNM. Based on 28 nm CMOS technology, experimental results demonstrate that the energy consumption of this PIC is 0.22 pJ/bit, while data movement efficiency improves by nearly 92% compared to traditional DRAM operations.

Virtual machines (VMs) impose increasing memory demands, exposing the capacity and cost limitations of traditional DRAM only memory architectures. To address this problem, heterogeneous DRAM+CXL tiered memory management systems have emerged as a promising solution. However, in virtualization environments, the semantic gap between guest and host abstraction layers, coupled with dynamic workload behaviors, hinders precise page tracking, classification, and efficient page migration across memory tiers.

Deep learning becomes increasingly popular, and its main workload is Sparse-Sparse Matrix Multiplication (SpMSpM). Most SpMSpM accelerators usually only support a single dataflow. Different dataflows have different performance in different computing environments. Therefore, the single-dataflow accelerator cannot maintain the highest performance in all environments. Compared with single-dataflow accelerators, multi-dataflow accelerators provide flexible options for different workloads and improve the overall performance. Flexagon, Sparm, and SPADA are state-of-the-art multi-dataflow accelerators. However, the computation process of Flexagon and Sparm is not fully pipelined, and SPADA cannot support inner product dataflow. Additionally, Flexagon, Sparm, and SPADA cannot switch dataflows quickly and accurately. Inspired by these observations, we present SpMARD, a SpMSpM accelerator with reconfigurable dataflow. The computation process of SpMARD is fully pipelined, and SpMARD can support six dataflow variants simultaneously. Through the design of a Two-stage Pipeline Adder Network (TPAN) and a Position-based Psum Array (PPA), SpMARD can execute element-level merging, which can hide the merging overhead. Through the quantitative analysis of dataflows, we implement a Dataflow Switcher (DSwitcher), which can switch dataflows more efficiently. For the SpMSpM workload, the performance (GOPS) of the SpMARD we proposed is 1.27 times that of Flexagon, 1.18 times that of Sparm, and 1.22 times that of SPADA.

Design Space Exploration (DSE) aims at identifying Pareto optimal synthesis configurations. Previous works require microarchitecture samples with key labels, including power and clock cycles, to train their models. However, as the chip design space expands rapidly, the cost of sampling the design space has significantly increased, due to the growing number of samples and time-consuming Very Large Scale Integration (VLSI) implementation flow. Recent advancements in Large Language Models (LLMs) have demonstrated their remarkable power in zero-shot learning tasks, presenting an innovative strategy for accomplishing DSE. Hence, this article presents ChatDSE, a zero-shot framework for DSE that is powered by the advanced capabilities of the LLM GPT4.0. Firstly, this framework analyzes the nature of the target microarchitecture and generates a corresponding system context to provide the prior knowledge of the microarchitecture. Secondly, a proposed sampling algorithm, PriorDC, identifies the most representative samples with pseudo labels. One of these samples is chosen as a baseline, whose power and clock cycles labels are set as 1, and the remaining sample labels are obtained by chatting with GPT4.0. Finally, ChatDSE engages in a dialogue with GPT4.0 to estimate the power and clock cycles of designs within the space, ultimately identifying the Pareto optimal design set. In the DSE for the RISC-V Berkeley Out-of-Order Machine (BOOM), experimental results show that ChatDSE is capable of identifying optimal designs and accelerates the exploration process by 574 times when compared to the state-of-the-art DSE methodologies.

Computing-in-memory (CIM) is a promising technique for hardware acceleration of neural networks (NNs) with high performance and efficiency. However, conventional dense mapping scheme cannot well support the compression and optimization of irregular sparse NNs. In this paper, we propose a heuristic and greedy weight remapping scheme for irregular sparse neural networks implemented on CIM accelerator in edge AI applications. The genetic algorithm (GA) is proposed for the first time to be utilized in the column shuffle for sparse weight remapping. Combined with the granularity exploration of the CIM, the proportion of the compressible all-zero rows increase remarkably. A greedy algorithm is then employed to planarize the unevenly compressed units, thus to improve the storage utilization of the crossbar. For hardware optimization, the pipeline is customized with a zero-skipping circuit to leverage the bit-level activation sparsity at runtime. Our results show that the proposed remapping scheme achieves 70%-94% utilization rate of the sparsity, and an average of $1.3 \times$ increment compared with the naive compression. The cooptimized CIM achieves $3-7.6 \times$ speedup and $2.1- 4.8 \times$ energy efficiency, compared with the baseline for dense NNs.

The use of Light Detection and Ranging (LiDAR) for sensing has continuously improved the precision and performance of autonomous driving. At the same time, the large number of high-precision point clouds generated by LiDAR require real-time processing, and the range search is the key part of the processing pipeline. Content-Addressable Memory (CAM) has proven its efficiency for search tasks on switches and routers, but so far, there is still a lack of exploration on its application in point cloud range search. In this work, we propose CAMPER, a CAM-centered accelerator, aiming to explore the potential of CAM for point cloud range search. We developed a ripple comparison 13T (RC-13T) CAM cell for distance comparison, designed a spatial approximation search algorithm based on Chebyshev distance, and discussed the flexibility and scalability of the architecture. The results show that in the range search task of 64k@64k points, CAMPER achieves a latency of 0.83ms and a power consumption of 114.6mW. Compared with GPU, the throughput is increased by 10.4×; compared with SOTA accelerator, the energy efficiency is increased by about 228×.

3D Gaussian Splatting (3DGS) has recently emerged as a promising technique in the realms of 3D vision and robotics. Its capacity for rapid rendering and high-fidelity reconstruction makes it an attractive candidate for integration into Simultaneous Localization and Mapping (SLAM) systems. However, existing 3DGS-based SLAM systems still suffer from inadequate tracking throughput due to tremendous recursion in volume rendering and irregular memory access for gradient backpropagation. To address these challenges, this paper proposes GauSPU, an algorithm-hardware co-designed accelerator for supporting real-time 3DGS-based SLAM. On the algorithm side, we present a sparse-tile-sampling (STS) method for efficient pose tracking. The STS focuses on informative image regions, discarding the rest to alleviate computational workload while maintaining accuracy. At the hardware level, we make twofold efforts. Firstly, we design a sparsity-adaptive ray recursion unit (SA-RRU) to accelerate volume rendering by leveraging irregular spatial sparsity. The SA-RRU introduces a sub-tile-wise execution pattern and a Morton-based thread allocation scheme to optimize sparsity utilization. Additionally, a sparsity-aware task dispatcher ensures efficient fine-grained task scheduling. Secondly, we propose a memory-access-relaxed backpropagation engine (MAR-BE) for efficient gradient aggregation. It comprises a gradient buffer unit (GBU) for coalescing partial gradients and a pose backward unit (PBU) for pipeline-fused backpropagation, collaboratively eliminating the costly atomic operations. Sufficient experiments demonstrate that, through the integration of GauSPU and GPU, the system achieves a throughput of 33.6 FPS for real-time pose tracking in 3DGS-SLAM, presenting a significant$63.9\times$improvement in energy efficiency compared to the RTX3090 baseline.

Deep learning has become a highly popular research field, and previously deep learning algorithms ran primarily on CPUs and GPUs. However, with the rapid development of deep learning, it was discovered that existing processors could not meet the specific large-scale computing requirements of deep learning, and custom deep learning accelerators have become popular. The majority of the primary workloads in deep learning are general matrix-matrix multiplications (GEMMs), and emerging GEMMs are highly sparse and irregular. The TPU and SIGMA are typical GEMM accelerators in recent years, but the TPU does not support sparsity, and both the TPU and SIGMA have insufficient utilization rates of the Processing Element (PE). We design and implement SparGD, a sparse GEMM accelerator with dynamic dataflow. SparGD has specific PE structures, flexible distribution networks and reduction networks, and a simple dataflow switching module. When running sparse and irregular GEMMs, SparGD can maintain high PE utilization while utilizing sparsity, and can switch to the optimal dataflow according to the computing environment. For sparse, irregular GEMMs, our experimental results show that SparGD outperforms systolic arrays by 30 times and SIGMA by 3.6 times.

Neural radiance field (NeRF) has proved to be promising in augmented/virtual-reality applications. However, the deployment of NeRF on edge devices suffers from inadequate throughput due to redundant ray sampling and congested memory access. To address these challenges, this article proposes Hi-NeRF, a multirendering-core accelerator for efficient edge NeRF rendering. On the architecture level, a hierarchical empty space skipping (HESS) scheme is adopted, which efficiently locates the effective samples with fewer skipping steps and thus accelerates the ray marching process. Furthermore, to alleviate the memory access bottleneck, a vertex-interleaved mapping (VIM) method that eliminates memory bank conflicts is also proposed. On the hardware level, ineffective sample filters (ISFs) and voxel access filters (VCFs) are introduced to further exploit spatial sparsity and data locality at run-time. The experimental results show that our work achieves$2.67\times $rendering throughput and$11.2\times $energy efficiency compared to a SOTA NeRF rendering accelerator. The energy efficiency can be improved by$561\times $compared to a commercial GPU.

High-quality random numbers are very critical to many fields such as cryptography, finance, and scientific simulation, which calls for the design of reliable true random number generators (TRNGs). Limited by entropy source, throughput, reliability, and system integration, existing TRNG designs are difficult to be deployed in real computing systems to greatly accelerate target applications. This study proposes a TRNG circuit named resilient high-speed (RHS)-TRNG based on spin-transfer torque magnetic tunnel junction (STT-MTJ). RHS-TRNG generates resilient and high-speed random bit sequences exploiting the stochastic switching characteristics of STT-MTJ. By circuit/system codesign, we integrate RHS-TRNG into a reduced instruction set computer-V (RISC-V) processor as an acceleration component, which is driven by customized random number generation instructions. Our experimental results show that a single cell of RHS-TRNG has a random bit generation speed of up to 303 Mb/s, which is the highest among existing MTJ-based TRNGs. Higher throughput can be achieved by exploiting cell-level parallelism. RHS-TRNG also shows strong resilience against PVT variations thanks to our designs using bidirectional switching currents and dual generator units. In addition, our system evaluation results using gem5 simulator suggest that the system equipped with RHS-TRNG can achieve 3.4–$12\times $higher performance in speeding up option pricing programs than software implementations of random number generation.

Electronics employed in modern safety-critical systems require severe qualification during the manufacturing process and in the field, to prevent fault effects from manifesting themselves as critical failures during mission operations. Traditional fault models are not sufficient anymore to guarantee the required quality levels for chips utilized in mission-critical applications. The research community and industry have been investigating new test approaches such as device-aware test, cell-aware test, path-delay test, and even test methodologies based on the analysis of manufacturing data to move the scope from OPPM to OPPB. This special session presents four contributions, from academic researchers and industry professionals, to enable better chip quality. We present results on various activities towards this objective, including device-aware test, software-based self-test, and memory test.

STT-MRAM has long been a promising non-volatile memory solution for the embedded application space owing to its attractive characteristics such as non-volatility, low leakage, high endurance, and scalability. However, the operating requirements for high-performance computing (HPC) and low power (LP) applications involve different challenges. This paper addresses different aspects of STT-MRAM; it will cover state-of-the-art, some new results and future challenges related to technology, design and test. While STT-MRAM devices have shown encouraging performance metrics at device-level, a key challenge has been achieving backend-of-line (BEOL) CMOS compatibility, while retaining the benefits of low power operation. Scaling demands to improve data densities have placed additional challenges in terms of addressing the impact of process-induced damage on device performance at CD < 100 nm. In addition, the paper discusses the design of reliable read mechanism considering the variability effects. Moreover, the failure of traditional fault modeling and test approaches in model STT-MRAM unique defects for appropriate test solutions is demonstrated in this paper based on silicon data.

Resistive RAM (RRAM) is a promising technology to replace traditional technologies such as Flash, because of its low energy consumption, CMOS compatibility, and high density. Many companies are prototyping this technology to validate its potential. Bringing this technology to the market requires high-quality tests to ensure customer satisfaction. Hence, it is of great importance to deeply understand manufacturing defects and accurately model them to develop optimal tests. This paper presents a holistic framework for defect and fault modeling that enables the development of optimal tests for RRAMs. An overview and classification of RRAM manufacturing defects are provided. Defects in contacts and interconnects are modeled as resistors. Unique RRAM defects, e.g., forming defects, require Device-Aware defect modeling which incorporates the defect’s impact on the device’s electric properties by adjusting the affected technology and electrical parameters. Additionally, a systematic approach to define the fault space is presented, followed by a methodology to validate this space. With this methodology, accurate fault modeling for contact, interconnect, and forming defects is performed and tests are developed. The tests are able to detect all faults in a time-efficient manner, thereby proving the effectiveness of the framework. Finally, an outlook on future RRAM testing is presented.

Understanding defects in magnetic tunnel junctions (MTJs) and their faulty behaviors are paramount for developing high-quality test solutions for STT-MRAM. This article applies the advanced device-aware test to intermediate (IM) state defects in MTJ devices based on silicon measurements and circuit simulations. An IM state manifests itself as an abnormal third resistive state, which differs from the two bi-stable states of MTJ. We performed silicon measurements on MTJ devices with diameter ranging from 60 nm to 120 nm; the results show that the occurrence probability of IM state strongly depends on the switching direction, device size, and bias voltage. We demonstrate that the conventional resistor-based fault modeling and test approach fails to appropriately model and test such a defect. Therefore, device-aware test is applied. We first physically model the defect and incorporate it into a Verilog-A MTJ compact model and calibrate it with silicon data. Thereafter, this model is used for a systematic fault analysis based on circuit simulations to obtain accurate and realistic faults in a pre-defined fault space. Our simulation results show that an IM state defect leads to intermittent write transition faults. Finally, we propose and implement a device-aware test solution to detect the IM state defect.

The popularity of perpendicular magnetic tunnel junction (pMTJ)-based spin-transfer torque magnetic random access memories (STT-MRAMs) is growing very fast. The performance of such memories is very sensitive to magnetic fields, including both internal and external ones. This article presents a magnetic-field-aware compact model of pMTJ, named the MFA-magnetic tunnel junction (MTJ) model, for magnetic/electrical co-simulation of MTJ/CMOS circuits. Magnetic measurement data of MTJ devices, with diameters ranging from 35 to 175 nm, are used to calibrate an in-house magnetic coupling model. This model is subsequently integrated into our developed compact pMTJ model, which is implemented in Verilog-A. The superiority of the proposed MFA-MTJ model for device/circuit co-design of STT-MRAM is demonstrated by simulating a single pMTJ as well as STT-MRAM full circuits. The design space is explored under PVT variations and various configurations of magnetic fields.

Understanding the defects in magnetic tunnel junctions (MTJs) and their faulty behaviors are paramount for developing high-quality tests for STT-MRAM. This paper characterizes and models intermediate (IM) state defects in MTJs; IM state manifests itself as an abnormal third resistive state, apart from the two bi-stable states of MTJ. We performed silicon measurements on MTJ devices with diameter ranging from 60 nm to 120 nm; the results reveal that the occurrence probability of IM state strongly depends on the switching direction, device size, and applied bias voltage. To test such defect, appropriate fault models are needed. Therefore, we use the advanced device-aware modeling approach, where we first physically model the defect and incorporate it into a Verilog-A MTJ compact model and calibrate it with silicon data. Thereafter, we use a systematic fault analysis to accurately validate a theoretically predefined fault space and derive realistic fault models. Our simulation results show that the IM state defect causes intermittent write transition faults. This paper also demonstrates that the conventional resistor-based fault modeling and test approach fails in appropriately modeling IM defects, and hence incapable of detecting such defects.

STT-MRAM is one of the most promising emerging non-volatile memory technologies. As its mass production and deployment in industry is around the corner, high-quality yet cost-efficient manufacturing test solutions are crucial to ensure the required quality of products being shipped to end customers. This paper focuses on STT-MRAM testing, covering three abstraction levels: manufacturing defects, fault models, and test solutions. We first survey STT-MRAM manufacturing defect space and apply the conventional resistor-based test approach to develop test solutions. We then demonstrate with silicon measurements that this approach fails to appropriately model and test defects in STT-MRAM devices: magnetic tunnel junctions (MTJs), although it is qualified for interconnect/contact defects. Therefore, we propose a new test approach: device-aware test (DAT) to specifically target device-internal defects. We apply DAT to three key types of MTJ defects: pinhole, synthetic anti-ferromagnet flip, and intermediate state defects. After developing accurate defect models and calibrating them with silicon data, we perform comprehensive fault analyses based on SPICE circuit simulations to derive accurate and realistic fault models. Some STT-MRAM unique faults are identified, including both permanent faults and intermittent faults. Based on the obtained fault models, high-quality test solutions are proposed. Additionally, this paper also proposes a magnetic coupling model and a magnetic-field-aware compact MTJ model for fast and robust STT-MRAM designs.

Manufacturing defects can cause hard-to-detect (HTD) faults in fin field-effect transistor (FinFET) static random access memories (SRAMs). Detection of these faults, such as random read outputs and out-of-spec parametric deviations, is essential when testing FinFET SRAMs. Undetected HTD faults result in test escapes, which lead to early in-field failures. This article presents a detailed analysis of HTD faults in FinFET SRAMs by exploring their sensitization and discussing solutions to improve HTD fault coverage during manufacturing testing. We first define the fault space for SRAMs and classify all faults in the space. Following this, we perform a systematic fault analysis based on injecting resistive defects in a memory cell, inspecting its behavior, and identifying HTD faults. Furthermore, we survey existing test solutions and discuss their HTD fault coverage and limitations. Based on our analysis, it is clear that no single test solution can fully detect all HTD faults, thus leading to test escapes. Hence, there is a need for new and more efficient test solutions. Improved detection of HTD faults could be achieved by using parametric test solutions, proposing solutions that cover yet-untargeted HTD faults, combining multiple test approaches into a single solution, and further exploring stress conditions. These new approaches would reduce test escapes and therefore improve the quality of FinFET SRAMs.

Manufacturing defects can cause faults in FinFET SRAMs. Of them, easy-to-detect (ETD) faults always cause incorrect behavior, and therefore are easily detected by applying sequences of write and read operations. However, hard-to-detect (HTD) faults may not cause incorrect behavior, only parametric deviations. Detection of these faults is of major importance as they may lead to test escapes. This paper proposes a new design-for-testability (DFT) scheme for FinFET SRAMs to detect such faults by creating a mismatch in the sense amplifier (SA). This mismatch, combined with the defect in the cell, will incorrectly bias the SA and cause incorrect read outputs. Furthermore, post-silicon calibration schemes can be used to avoid over-testing or test escapes caused by process variation effects. Compared to the state of the art, this scheme introduces negligible overheads in area and test time while it significantly improves fault coverage and reduces the number of test escapes.

As a unique mechanism for MRAMs, magnetic coupling needs to be accounted for when designing memory arrays. This paper models both intra- and inter-cell magnetic coupling analytically for STT-MRAMs and investigates their impact on the write performance and retention of MTJ devices, which are the data-storing elements of STT-MRAMs. We present magnetic measurement data of MTJ devices with diameters ranging from 35 nm to 175 nm, which we use to calibrate our intra-cell magnetic coupling model. Subsequently, we extrapolate this model to study inter-cell magnetic coupling in memory arrays. We propose the inter-cell magnetic coupling factor Ψ to indicate coupling strength. Our simulation results show that Ψ≈2% maximizes the array density under the constraint that the magnetic coupling has negligible impact on the device's performance. Higher array densities show significant variations in average switching time, especially at low switching voltages, caused by inter-cell magnetic coupling, and dependent on the data pattern in the cell's neighborhood. We also observe a marginal degradation of the data retention time under the influence of inter-cell magnetic coupling.

This paper introduces a new test approach: device-aware test (DAT) for emerging memory technologies such as MRAM, RRAM, and PCM. The DAT approach enables accurate models of device defects to obtain realistic fault models, which are used to develop high-quality and optimized test solutions. This is demonstrated by an application of DAT to pinhole defects in STT-MRAMs and forming defects in RRAMs.

Understanding the manufacturing defects in magnetic tunnel junctions (MTJs), which are the data-storing elements in STT-MRAMs, and their resultant faulty behaviors are crucial for developing high-quality test solutions. This paper introduces a new type of MTJ defect: synthetic anti-ferromagnet flip (SAFF) defect, wherein the magnetization in both the hard layer and reference layer of MTJ devices undergoes an unintended flip to the opposite direction. Both magnetic and electrical measurement data of SAFF defect in fabricated MTJ devices is presented; it shows that such a defect reverses the polarity of stray field at the free layer of MTJ, while it has no electrical impact on the single isolated device. The paper also demonstrates that using the conventional fault modeling and test approach fails to appropriately model and test such a defect. Therefore device-aware fault modeling and test approach is used. It first physically models the defect and incorporate it into a Verilog-A MTJ compact model, which is afterwards calibrated with silicon data. The model is thereafter used for fault analysis and modeling within an STT-MRAM array; simulation results show that a SAFF defect may lead to an intermittent Passive Neighborhood Pattern Sensitive Fault (PNPSF1i) when all neighboring cells are in logic `1' state. Finally, test solutions for such fault are discussed.

Emerging memristor-based architectures are promising for data-intensive applications as these can enhance the computation efficiency, solve the data transfer bottleneck and at the same time deliver high energy efficiency using their normally-off/instant-on attributes. However, their storing devices are more susceptible to manufacturing defects compared to the traditional memory technologies because they are fabricated with new materials and require different manufacturing processes. Hence, in order to ensure correct functionalities for these technologies, it is necessary to have accurate fault modeling as well as proper test methodologies with high test coverage. In this paper, we propose technology specific cell-level defect modeling, accurate fault analysis and yield improvement solutions for memristor-based memory as well as Computation-In-Memory (CIM) architectures. Our overall contributions cover three abstraction levels, namely, device, architecture and system. First, we propose a device-aware test methodology in which we have introduced a key device-level characteristic to develop accurate defect model. Second, we demonstrate a yield analysis framework for memristor arrays considering reliability and permanent faults due to parametric variations and explore fault-tolerant solutions. Third, a lightweight on-line test and repair schemes is proposed for emerging CIM devices in machine learning applications.

The STT-MRAM manufacturing process involves not only traditional CMOS process steps, but also the integration of magnetic tunnel junction (MTJ) devices, the data-storing elements. This paper demonstrates a paradigm shift in fault modeling for STT-MRAMs by performing defect modeling and fault analysis for MTJ pinhole defects which are seen as a key type of STT-MRAM manufacturing defects. A Verilog-A compact model for defect-free MTJ devices is built and calibrated with electrical measurements on actual MTJ wafers. MTJs with a pinhole defect are extensively characterized, both during manufacturing test (t=0) and in the field (t>0), and the data is used to extend our defect-free MTJ compact model to include parameterized pinhole defects. The model is then used to perform single-cell static fault analysis and this shows not only what kind of faults can occur in an STT-MRAM, but also that the conventional fault modeling approach based on linear resistors cannot catch such behavior.

This paper proposes a new test approach that goes beyond cell-aware test, i.e., device-aware test. The approach consists of three steps: defect modeling, fault modeling, and test/DfT development. The defect modeling does not assume that a defect in a device (or a cell) can be modeled electrically as a linear resistor (as the traditional approach suggests), but it rather incorporates the impact of the physical defect on the technology parameters of the device and thereafter on its electrical parameters. Once the defective electrical model is defined, a systematic fault analysis (based on fault simulation) is performed to derive appropriate fault models and subsequently test solutions. The approach is demonstrated using two memory technologies: resistive random access memory (RRAM) and spin-transfer torque magnetic random access memory (STT-MRAM). The results show that the proposed approach is able to sensitize faults for defects that are not detected with the traditional approach, meaning that the latter cannot lead to high-quality test solutions as required for a defective part per billion (DPPB) level. The new approach clearly sets up a turning point in testing for at least the considered two emerging memory technologies.

Spin-transfer-torque magnetic RAM (STT-MRAM) is one of the most promising emerging memory technologies. As various manufacturing vendors make significant efforts to push it to the market, appropriate STT-MRAM testing is of great importance. In this paper, we demonstrate that conventional STT-MRAM defect modeling, which is based on linear resistors, is too pessimistic in representing the real nature of physical defects. It may result in incorrect fault models, which in turn can lead to low-quality test solutions. In addition, we propose a generic defect modeling methodology which captures the nonlinear behavior of STT-MRAM defects accurately; a defect is modeled by adjusting the affected STT-MRAM technology parameters. The methodology is illustrated by two examples, namely a pinhole defect and a sidewall redeposition defect, which are simulated for accurate fault modeling. In case of a pinhole defect, the STT-MRAM suffers from a fast transition between magnetic tunnel junction (MTJ) states with increased write current, making the MTJ more vulnerable to breakdown. However, with the conventional linear resistor as defect model the memory shows a slow transition or even a transition failure. Similarly, a sidewall redeposition defect causes a fast transition without current elevation, which is not observed when using the conventional approach.