17 March 2023 in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
Compute-in-Memory (CIM) accelerator has become a popular solution to achieve high energy efficiency for deep learning applications in edge devices. Recent works have demonstrated CIM macros using non-volatile memories (STT-MRAM, RRAM) to take advantages of their non-volatility and high density. However, effective computation dynamic range is far lower than their SRAM-CIM counterparts due to low device ON/OFF ratio. In this work, we combine a non-volatile memory based on a voltage-controlled magnetic tunneling junction (VC-MTJ) device, called voltage-controlled MRAM or VC-MRAM, and accurate switched-capacitor based CIM using a novel in-situ Magnetic-to-Digital Converter (MDC). The VC-MTJ device has demonstrated 10× lower write energy and switching time compared to STT-MRAM device and has comparable density, read energy and read latency. The in-situ MDCs embedded inside each VC-MRAM row convert magnetically stored weight information to CMOS logic levels and enable switched-capacitor based multiply–accumulate (MAC) operation with accuracy comparable to state-of-the-art SRAM-CIM. This paper describes the schematic and layout level design of a VC-MRAM CIM macro in 28nm. This is the first non-volatile CIM design to enable analog MAC computation with 256 parallel rows turned ON simultaneously without degradation in dynamic range (< 1 LSB). Detailed circuit simulations including experimentally validated VC-MTJ compact models show 1.5× higher energy efficiency and 2× higher density compared to state-of-the-art SRAM-based CIM.
29 September 2025 in IEEE Journal of Solid-State Circuits
Memory performance has emerged as a critical factor influencing both system speed and energy efficiency. However, conventional memory technologies such as embedded Flash (eFlash) and static RAM (SRAM) encounter significant scalability limitations beyond the 28-nm CMOS node. Among novel emerging memory technologies, spin-transfer-torque magnetic RAM (STT-MRAM) has gained prominence due to its high-speed operation, non-volatility, and superior endurance, positioning it as a promising candidate for storage-class and cache memory applications. Nevertheless, the energy-intensive write mechanism of STT-MRAM poses a fundamental limitation to its broader adoption. In this work, we present the first CMOS-integrated implementation of a novel memory technology called the magneto-electric RAM (MeRAM) that utilizes voltage-controlled magnetic anisotropy (VCMA) for write operations, achieving sub-nanosecond switching speeds and ultralow energy consumption. The integration process, associated challenges, and circuit solutions are thoroughly discussed, including the development of a unique memory access circuitry tailored for MeRAM. The experimental results from multiple MeRAM chips fabricated in a 180-nm CMOS process are provided, showcasing key device characteristics and memory performance metrics. This work demonstrates the first fully operational, CMOS-integrated MeRAM chips, achieving sub-1-ns switching time, 8.85-pJ/bit write energy, and 1.95-pJ/bit read energy in a 180-nm CMOS technology node.
27 August 2024 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Event-based cameras offer low latency and high dynamic range imaging data in a sparse format that is well-suited for high-speed object tracking. Processing this sparse data in the same way as traditional camera data requires a great deal of unnecessary computation, making it difficult to take advantage of the high effective frame rate for real-time processing. In this work, we propose an accelerator for high-speed object tracking on event-based camera data. SCIMITAR combines digital in-memory stochastic computing, in-situ stochastic stream generation, and multiple optimizations for utilizing input sparsity. SCIMITAR provides unparalleled performance with latency and energy that scale with sparsity. We demonstrate SCIMITAR performance on an object tracking application using circuit-level simulations of custom-designed compute-in-memory macros and digital circuits. We achieve a frame processing rate of 2.6k frames per second with 100 Regions-Of-Interest per frame and equivalent or better than state-of-the-art tracking accuracy. The accelerator achieves a peak throughput of 71 TOP/S and energy efficiency of 733 to 1702 TOP/S/W demonstrated on a range of event-based vision datasets, which is 5× higher than other compute-in-memory solutions.
1 January 2024 in IEEE Transactions on Circuits and Systems I: Regular Papers
The filtering-by-aliasing (FA) receivers have demonstrated sharp analog finite impulse response (FIR) filtering by combining periodically time-varying (PTV) circuit elements with uniform sampling. Seen at the sampled output, the impulse response of an FA receiver can be controlled by the PTV resistor’s resistance that varies over time, which realizes very sharp filtering with 50–70-dB stopband rejection. However, the finite stopband rejection and the inherent sampling in FA result in unwanted in-band residual blocker aliases, which cannot be suppressed further by downstream linear time-invariant (LTI) stages, unlike in a conventional receiver, and the overall blocker rejection may be insufficient in some applications. This paper describes a digital residual alias cancellation technique tailored for the FA receivers. By using a second channel to capture the blocker and equalizing the blocker in the FA and the second channels with digital baseband filters, measurement results, with the cancellation algorithm implemented in MATLAB, show that the residual blocker aliases can be suppressed by another $\sim$ 15 dB, in addition to the FA analog filtering.
3 April 2026 in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
Compute in memory (CiM) accelerators perform matrix vector multiplications (MVMs) directly inside memory arrays, reducing data movement and improving both energy efficiency and throughput for AI workloads. To reduce the number of conversions, recent designs use multi-bit compute cells. Nevertheless, practical multi-bit CiM still faces a tension between accuracy, efficiency, and re-writeability, since multi-level NVM based designs suffer from nonlinearity and poor re-writeability, while multi-level activation based DRAM / SRAM macros are limited by mismatch and low accuracy. This work introduces a per-cell DAC based CiM macro that combines the density of multi-level NVM with fully re-writable DRAM weights to break the trade-off. Each bit cell embeds a compact 6-bit CTT based current mode DAC, calibrated in-situ through a write verify write loop, together with a 1T(1C) embedded DRAM. Post-layout simulations of a 576×256 macro in 22nm FDSOI project 49.9 8b-TOPS/W, and 8.96 8b-TOPS/mm2, while system level evaluation on CIFAR-10 and CIFAR-100 with ResNet-50 shows less than 0.2% accuracy loss relative to a digital baseline, without the need for any Hardware-aware Training.