Fusion of memristor and digital compute-in-memory processing for energy-efficient edge computing
The DOI matches its Crossref record; 38 of the 40 references it cites were found in a registry or an index.
Abstract
Artificial intelligence (AI) edge devices prefer employing high-capacity nonvolatile compute-in-memory (CIM) to achieve high energy efficiency and rapid wakeup-to-response with sufficient accuracy. Most previous works are based on either memristor-based CIMs, which suffer from accuracy loss and do not support training as a result of limited endurance, or digital static random-access memory (SRAM)-based CIMs, which suffer from large area requirements and volatile storage. We report an AI edge processor that uses a memristor-SRAM CIM-fusion scheme to simultaneously exploit the high accuracy of the digital SRAM CIM and the high energy-efficiency and storage density of the resistive random-access memory memristor CIM. This also enables adaptive local training to accommodate personalized characterization and user environment. The fusion processor achieved high CIM capacity, short wakeup-to-response latency (392 microseconds), high peak energy efficiency (77.64 teraoperations per second per watt), and robust accuracy (<0.5% accuracy loss). This work demonstrates that memristor technology has moved beyond in-lab development stages and now has manufacturability for AI edge processors.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
In-memory computing with resistive switching devices
Ielmini, D., & Wong, H.-S. P. (2018) · Nature Electronics · doi:10.1038/s41928-018-0092-2 · Cited by 2,298
Fully hardware-implemented memristor convolutional neural network
Yao, P., Wu, H., Gao, B., Tang, J., Zhang, Q., Zhang, W., Yang, J. J., & Qian, H. (2020) · Nature · doi:10.1038/s41586-020-1942-4 · Cited by 2,179
Analogue signal and image processing with large memristor crossbars
Li, C., Hu, M., Li, Y., Jiang, H., Ge, N., Montgomery, E., Zhang, J., Song, W., Dávila, N., Graves, C. E., et al. (2017) · Nature Electronics · doi:10.1038/s41928-017-0002-z · Cited by 1,292
Equivalent-accuracy accelerated neural-network training using analogue memory
Ambrogio, S., Narayanan, P., Tsai, H., Shelby, R. M., Boybat, I., di Nolfo, C., Sidler, S., Giordano, M., Bodini, M., Farinha, N. C. P., et al. (2018) · Nature · doi:10.1038/s41586-018-0180-5 · Cited by 1,151
A compute-in-memory chip based on resistive random-access memory
Wan, W., Kubendran, R., Schaefer, C., Eryilmaz, S. B., Zhang, W., Wu, D., Deiss, S., Raina, P., Qian, H., Gao, B., et al. (2022) · Nature · doi:10.1038/s41586-022-04992-8 · Cited by 943
Neuromorphic computing with multi-memristive synapses
Boybat, I., Le Gallo, M., Nandakumar, S. R., Moraitis, T., Parnell, T., Tuma, T., Rajendran, B., Leblebici, Y., Sebastian, A., & Eleftheriou, E. (2018) · Nature Communications · doi:10.1038/s41467-018-04933-y · Cited by 899
A fully integrated reprogrammable memristor–CMOS system for efficient multiply–accumulate operations
Cai, F., Correll, J. M., Lee, S. H., Lim, Y., Bothra, V., Zhang, Z., Flynn, M. P., & Lu, W. D. (2019) · Nature Electronics · doi:10.1038/s41928-019-0270-x · Cited by 728
A crossbar array of magnetoresistive memory devices for in-memory computing
Jung, S., Lee, H., Myung, S., Kim, H., Yoon, S. K., Kwon, S.-W., Ju, Y., Kim, M., Yi, W., Han, S., et al. (2022) · Nature · doi:10.1038/s41586-021-04196-6 · Cited by 644
Mixed-precision in-memory computing
Le Gallo, M., Sebastian, A., Mathis, R., Manica, M., Giefers, H., Tuma, T., Bekas, C., Curioni, A., & Eleftheriou, E. (2018) · Nature Electronics · doi:10.1038/s41928-018-0054-8 · Cited by 485
Chih, Y.-D., Lee, P.-H., Fujiwara, H., Shih, Y.-C., Lee, C.-F., Naous, R., Chen, Y.-L., Lo, C.-P., Lu, C.-H., Mori, H., et al. (2021) · 2021 IEEE International Solid- State Circuits Conference (ISSCC) · doi:10.1109/isscc42613.2021.9365766 · Cited by 384
Self-citation
Cited by
Items 101–126 of 126
SALMON: Self-Adaptive Learning Model on Neuromorphic Hardware
Cho, Y. W., Lee, S., & Kim, S. (2026) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2026.3687785
“To address the limited conductance states and D2D variations in analog memory devices, analog-digital hybrid weight methods also have been proposed [25], [26], [27], [28], [29].”
Bae, J., Han, J. H., Kim, T., Won, J., Park, S., Kim, G. W., Kim, S., Choi, H., Kim, T., Lee, H., et al. (2026) · Advanced Materials · doi:10.1002/adma.202523670
Neuromorphic Denoising with Fully Analog Memristive In‐Memory Computing
Shi, D., Zhang, T., Li, Y., Tao, Y., Yan, B. Y., & Yang, Y. (2026) · Advanced Intelligent Systems · doi:10.1002/aisy.70419 · Cited by 1
Li, Z., Yang, S., Xiang, L., Gong, X., Chen, X., Hu, W., Lin, Z., Wu, X., & Peng, C. (2026) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2026.3696148
“Among various non-volatile memory (NVM) candidates [5], [6], [7], [8], [9], the Ferroelectric Field-E ff ect Transistor (FeFET) stands out due to its high-speed voltage-controlled operation, zero standby leakage, and exceptional CMOS compatibility in advanced technology nodes (e.g., 28nm) [10],…”
Zheng, L., Bavani, A. M., Chen, M., Du, S., Hsin, T.-Y., Khwa, W.-S., Wu, P.-S., Lele, A. S., Zhang, B., Crafton, B., et al. (2026) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2026.3694285 · Cited by 1
“…random access memory (RRAM)-CIM macros and systems, some combined with near-SRAM [24] or in-SRAM compute macros [25], [26], are demonstrated for applications ranging from edge inference [25], [27], [28], [29] and fine-tuning [26], to associative one-shot learning [30] and event-driven tracking [24].”
Self-citationIn-Memory Computing Enabled Deep MIMO Detection to Support Ultra-Low-Latency Communications
Ding, T., Zeng, Q., & Huang, K. (2026) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2026.3699403 · Cited by 1
“This motivates us to propose leveraging in-memory computing, which is an analog computing technology that integrates memory and computation within memristor circuits, to perform the intensive matrix-vector multiplication (MVM) operations inherent in deep MIMO detection at the nanosecond scale.”
Jiang, P., Wang, L., Fang, Y., Wang, Y., Huang, J.-M., Zhu, M., Miao, X., & Wang, X. (2026) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2026.3703379
“To reduce data movement during model inference, researchers have proposed non-volatile memory (NVM)-based analog CIM architectures that perform in situ matrix–vector multiplications (MVMs) [26], [27], [28].”
Li, R. K., Jiang, X., Zhao, X., Zhang, H., Xu, Q., & Wang, G. (2026) · Coatings · doi:10.3390/coatings16060715
Bao, H., Xu, K., Xue, Y., et al. (2026) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas66217.2026.11562913
“…is one representing the solutions of the problem [3], [4]. on novel non-volatile memories such as memristors, can achieve matrix-vector multiplications (MVMs) with high parallelism according to Ohm’s law and Kirchhoff’s current law, thus efficiently accelerating the iteration of the HNN [5-7].”
Zhou, Z., Fu, Y., Li, J., Li, Y., & Miao, X. (2026) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas66217.2026.11561964
“Among emerging solutions, resistive random access memory (RRAM) computing-in-memory (CIM) architectures offer a promising paradigm for energy-efficient energy-efficient AI inference [3], [4], [5], as illustrated in Fig.”
A Data-Driven Framework for Modeling Non-Idealities in Analog In-Memory Computing Hardware
Vasilopoulos, A., Kersting, B., Büchel, J., Sebastian, A., & Gallo, M. L. (2026) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2026.3705427
“Multiple AIMC prototypes have demonstrated this principle in practice, from large-scale systems [7], [8], [9], [10], [11] to smaller macro-level demonstrations [12], [13], [14], collectively establishing AIMC as a viable technology for e ffi cient deep learning (DL) inference.”
Self-rectifying organic memristor based on PAA/PEDOT:PSS heterojunction for neuromorphic computing
Ma, X., Tang, X., Shi, J., He, N., Ha, S., Sun, W., Song, X., Cai, G., & Zhao, J. (2026) · Applied Physics Letters · doi:10.1063/5.0334702
You, D.-Q., Khwa, W.-S., Zhang, B., Wu, J.-J., Jhang, C.-J., Lin, G.-Y., Chen, P., Chiu, T.-C., Chen, F., Lee, A., et al. (2026) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2026.3705432
“…[19], [20], [21], [22] and near-memory-compute (nvNMC) [23], [24], is ideally suited to energy-constrained nonvolatile AI-edge processors, thanks to its integrated computation and high-capacity on-chip weight storage in power-o ff mode [25], [26], [27], [28], [29], [30], [31], [32], [33], [34].”
Self-citationLi, R., Li, J., Xu, C., Yao, P., Wu, D., Jie, L., He, Y., Qin, Q., Pan, S., Zhang, Q., et al. (2026) · IEEE Electron Device Letters · doi:10.1109/led.2026.3707870
“Compute-in-memory (CIM) based on resistive random-access memory (RRAM) has emerged as a promising approach for AI accelerators and neuromorphic hardware to address the memory wall and von Neumann bottleneck [1], [2], [3], [4].”
Edge artificial intelligence: Past advances, current paradigms, and emerging horizons
Ali, M. A., Dornaika, F., & Charafeddine, J. (2026) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2026.115635 · Cited by 2
Low‐Dimensional Perovskites for Neuromorphic Vision Computing
Li, D., Zhang, S., Xie, P., Su, L., & Ho, J. C. Y. (2026) · Advanced Materials · doi:10.1002/adma.74278 · Cited by 2
“Typical architectures employ analog computing based on physical principles such as Ohm’s Law, Kirchhoff’s Current Law, and charge sharing within memory arrays (Figure 17a ) [ 212 ] 17b ) [ 215 ].”
Seo, Y., Kim, D., Park, Y. B., et al. (2026) · Advanced Materials · doi:10.1002/adma.74743
Metal‐Filamentary Memristor‐Based Artificial Neurons for Embodied Neuromorphic Intelligence
Huang, T., Yuan, Z., & Hua, Q. (2026) · Rare Metals · doi:10.1002/rar2.70514
Ligand polarization-driven surface rigid confinement for reliable perovskite quantum dot memristors
Peng, Z., Wang, Z., Gan, Y., et al. (2026) · Chemical Engineering Journal · doi:10.1016/j.cej.2026.181307
Algorithm--Hardware Co-Design of Spiking Transformers for In-Memory Neuromorphic Edge Vision
Li, Y., Hu, W., Peng, H., Deng, H., Hu, M., Xiang, J., & Hao, R. (2026) · Proceedings of the International Conference on Parallel Processing · doi:10.1145/3832810.3832823
“Crucially, THT’s binary operating regime substantially relaxes read-precision requirements, significantly reducing the energy and area overhead of necessary peripheral circuits (e.g., sense amplifiers and calibration units) compared to conventional analog matrix multiplication [6, 21].”
InfluentialGong, J., Gong, F., Wang, X., Deng, M., Ni, Y., Ye, H., Wei, H., Han, S., & Wei, H. (2026) · Device · doi:10.1016/j.device.2026.101301
Du, W., & Li, M. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7493629
Memristive singular value decomposition
Ding, C., Liu, Z., Zhang, Y., Li, C., Tang, J., Gao, B., Yu, H., Wu, H., & Wong, N. (2026) · Nature Communications · doi:10.1038/s41467-026-76272-2 · Cited by 1
“Recent studies have actively explored memristor-based CIM implementations for accelerating deep learning tasks 29 – 37 .”
Zhang, J., Yan, J., He, N., et al. (2026) · Small · doi:10.1002/smll.76033
Li, S., Kong, C., Liu, Y., Wang, Z., Yang, C., Xu, Q., Guo, X., Zhou, X., Guo, H., Jiang, S., et al. (2026) · ACS Materials Letters · doi:10.1021/acsmaterialslett.6c00732
Metadata
- Authors
- Wen, T.-H., Hung, J.-M., Huang, W.-H., Jhang, C.-J., Lo, Y.-C., Hsu, H.-H., Ke, Z.-E., Chen, Y.-C., Chin, Y.-H., Su, C.-I., Khwa, W.-S., Lo, C.-C., Liu, R.-S., Hsieh, C.-C., Tang, K.-T., Ho, M.-S., Chou, C.-C., Chih, Y.-D., Chang, T.-Y. J., & Chang, M.-F.
- Affiliations
- National Tsing Hua University (TW)
- Taiwan Semiconductor Manufacturing Company (Taiwan) (TW)
- National Chung Hsing University (TW)
- Corresponding author
- Wen, Tai-Hao
- Hung, Je-Min
- Huang, Wei-Hsing
- Jhang, Chuan-Jia
- Lo, Yun-Chen
- Chang, Meng-Fan
- ORCID
- Venue
- Science
- Published
- 2024-04-19
- Volume
- 384
- Issue
- 6693
- Pages
- 325-332
- ISSN
- 0036-8075 (ISSN-L)
- 1095-9203
- OpenAlex
- W4394909859
- Cited by
- 131
- References
- 40
- Citations by year
- 2026: 63 · 2025: 57 · 2024: 11
- FWCI
- 13.50
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Science (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
- Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Low-power high-performance VLSI design
- Keywords
- memristors, in-memory computing, SRAM, resistive random access memory, energy efficiency, personalized machine learning, non-volatile memory, wake-up latency
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.