CONV-SRAM: An Energy-Efficient SRAM With In-Memory Dot-Product Computation for Low-Power Convolutional Neural Networks
Open accessThe DOI matches its Crossref record; 25 of the 29 references it cites were found in a registry or an index.
Abstract
This paper presents an energy-efficient static random access memory (SRAM) with embedded dot-product computation capability, for binary-weight convolutional neural networks. A 10T bit-cell-based SRAM array is used to store the 1-b filter weights. The array implements dot-product as a weighted average of the bitline voltages, which are proportional to the digital input values. Local integrating analog-to-digital converters compute the digital convolution outputs, corresponding to each filter. We have successfully demonstrated functionality (>98% accuracy) with the 10 000 test images in the MNIST hand-written digit recognition data set, using 6-b inputs/outputs. Compared to conventional full-digital implementations using small bitwidths, we achieve similar or better energy efficiency, by reducing data transfer, due to the highly parallel in-memory analog computations.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Among charge-based memories, SRAM cells (a bi-stable transistor structure typically made of two CMOS inverters connected back-to-back which retains a charge concentration, see Fig. 3a for an example of the structure of a crossbar array of 6T SRAM) have been widely used for VMM75–77.”
From Hardware implementation of memristor-based artificial neural networks (2024) · cites this paper as [76]
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 1×
“Convolutional neural networks (CNNs) provide state-of-theart results in a wide variety of AI/ML applications, ranging from image classification [3] to speech recognition [1].”
BackgroundGradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 5×
“However, for the MNIST dataset with LeNet-5 CNN, we estimate [22] those to have only small contributions to the overall energy-efficiency per MAV operation, due to the high parallelism supported by our in-memory approach.”
MethodInfluentialHinton, G., Deng, L., Yu, D., Dahl, G., Mohamed, A.-R., Jaitly, N., Senior, A., Vanhoucke, V., Nguyen, P., Sainath, T., et al. (2012) · IEEE Signal Processing Magazine · doi:10.1109/msp.2012.2205597 · Cited by 10,260
DeepFace: Closing the Gap to Human-Level Performance in Face Verification
Taigman, Y., Yang, M., Ranzato, M., & Wolf, L. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.220 · Cited by 6,542 · cited 1×
“Facial recognition [2] enables using our faces to get access to devices in a more intuitive manner, instead of traditional passcodes.”
BackgroundXNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks
Rastegari, M., Ordonez, V., Redmon, J., & Farhadi, A. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_32 · Cited by 4,408 · cited 2×
“[6] proposed a binary-weight-network (BWN), where the filter weights (wi s) can be trained to be +1/−1 (with a common scaling factor per 3-D filter: α).”
BackgroundMethodEfficient Processing of Deep Neural Networks: A Tutorial and Survey
Sze, V., Chen, Y.-H., Yang, T.-J., & Emer, J. S. (2017) · Proceedings of the IEEE · doi:10.1109/jproc.2017.2761740 · Cited by 4,215 · cited 1×
“In the conventional all-digital implementation of CNNs [4], [5], [9], with the memory and the processing elements being physically separate, reading wi s and the partial sums from the on-chip SRAMs lead to a lot of data movement per computation [10] and, hence, make them energy-hungry.”
BackgroundEyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural Networks
Chen, Y.-H., Krishna, T., Emer, J. S., & Sze, V. (2017) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2016.2616357 · Cited by 3,247 · cited 12×
“CNNs typically consist of a cascade of convolutional (CONV) and fully-connected (FC) layers (Fig.”
BackgroundMethodInfluentialBinaryConnect: Training Deep Neural Networks with binary weights during propagations
Courbariaux, M., Bengio, Y., & David, J.-P. (2015) · arXiv · doi:10.48550/arxiv.1511.00363 · Cited by 2,180 · cited 1×
“BWN performs better than binary connect [7], which does not incorporate the scaling factor of α per filter, and also binarized neural networks [8], where both weights and activations are constrained to ±1.”
Background1.1 Computing's energy problem (and what we can do about it)
Horowitz, M. (2014) · 2014 IEEE International Solid-State Circuits Conference Digest of Technical Papers (ISSCC) · doi:10.1109/isscc.2014.6757323 · Cited by 2,069 · cited 1×
“to access data from memory can be much higher than the energy needed for a compute operation with that data [11].”
BackgroundHubara, I., Soudry, D., & Yaniv, R. E. (2016) · arXiv · doi:10.48550/arxiv.1602.02505 · Cited by 1,667 · cited 1×
“BWN performs better than binary connect [7], which does not incorporate the scaling factor of α per filter, and also binarized neural networks [8], where both weights and activations are constrained to ±1.”
Background
Research works printed but not linked: 1
- whatmough, 2017, A 28 nm SoC with a 1.2 GHz 568 nJ/prediction sparse deep-neural-network engine with >0.1 timing error rate tolerance for IoT applications, IEEE Int Solid-State Circuits Conf (ISSCC) Dig Tech Papers, 242
Cited by
Items 101–200 of 414
Antolini, A., Zavalloni, F., Lico, A., Quqa, S., Greco, L., Mangia, M., Pareschi, F., Pasotti, M., & Scarselli, E. F. (2025) · Sensors · doi:10.3390/s25123618 · Cited by 22
Jeong, S., Kim, J., Jeong, M., & Lee, Y. (2022) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2022.3165352 · Cited by 22
An energy-efficient 10T SRAM in-memory computing macro for artificial intelligence edge processor
Rajput, A. K., Pattanaik, M., & Kaushal, G. (2023) · Memories - Materials Devices Circuits and Systems · doi:10.1016/j.memori.2023.100076 · Cited by 20
Memristive dynamics enabled neuromorphic computing systems
Yan, B., Yang, Y., & Huang, R. (2023) · Science China Information Sciences · doi:10.1007/s11432-023-3739-0 · Cited by 20
Lou, J., Freye, F., Lanius, C., & Gemmeke, T. (2023) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2023.3323205 · Cited by 22
“Various CIM-based analog techniques have been proposed [30], [31], [32], [33], [34], [35], [36], [37], [38].”
Gong, M., Cao, N., Chang, M., & Raychowdhury, A. (2020) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2020.3027801 · Cited by 21
“Examples include switched-capacitor matrix multipliers [2], time-domain solvers [3], binary convolution neural networks (CNN) [4] and binary output classifiers [5].”
BackgroundALPINE: Analog In-Memory Acceleration with Tight Processor Integration for Deep Learning
Klein, J. A. H., Boybat, I., Qureshi, Y. M., Dazzi, M., Levisse, A., Ansaloni, G., Zapater, M., Sebastian, A., & Atienza, D. (2022) · IEEE Transactions on Computers · doi:10.1109/tc.2022.3230285 · Cited by 27
“SRAM-based AIMC approaches are attractive owing to SRAM’s technological maturity and scalability to aggressive CMOS nodes [25], [26].”
MethodZhang, X., & Basu, A. (2022) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2022.3218573 · Cited by 22
Ananthanarayanan, S., Reniwal, B. S., & Upadhyay, A. (2023) · International Conference on VLSI Design · doi:10.1109/vlsid57277.2023.00035 · Cited by 18
Tunneling magnetoresistance materials and devices for neuromorphic computing
Yao, Y., Cheng, H., Zhang, B., Yin, J., Zhu, D., Cai, W., Li, S., & Zhao, W. S. (2023) · Materials Futures · doi:10.1088/2752-5724/ace3af · Cited by 25
“Traditional memory circuits, such as static random-access memories (SRAMs) [4, 5] or floating gate memories [6], can be used for this purpose, but this solution leads to synapses that are not compact.”
BackgroundHassanzadeh, P. (2021) · Life Sciences · doi:10.1016/j.lfs.2021.119272 · Cited by 18
Multi-Logic Sense Amplifier (MLSA) Design for In-Memory Computing
Chang, Y., & Chen, S.-H. (2023) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2023.3243434 · Cited by 17
“Digital Object Identifier 10.1109/JETCAS.2023.3243434 memory (DRAM) [2], [16], [19], static random access memory (SRAM) [3], [6], [18], and non-volatile memory (NVM) [4], [17], [20].”
BackgroundInfluentialKumari, T., Vardan, Y., Shambharkar, P. G., & Gandhi, Y. (2022) · International Conference on Computing Methodologies and Communication (ICCMC) · doi:10.1109/iccmc53470.2022.9753756 · Cited by 16
Partial-Sum Quantization for Near ADC-Less Compute-In-Memory Accelerators
Saxena, U., & Roy, K. (2023) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1109/islped58423.2023.10244291 · Cited by 20
Chen, Z., Jin, Q., Yu, Z., Wang, Y., & Yang, K. (2022) · Custom Integrated Circuits Conference · doi:10.1109/cicc53496.2022.9772826 · Cited by 18
Reconfigurable 10T SRAM for Energy-Efficient CAM Operation and In-Memory Computing
Zhang, Z., Chen, Z., Wang, J., Xie, G., & Liu, G. (2025) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2025.3526973 · Cited by 15
“For instance, Biswas and Chandrakasan [9] and Jiang et al. [10] have realized analog MAC using charge-sharing and capacitive coupling schemes, respectively.”
Rajput, A. K., Pattanaik, M., & Kaushal, G. (2022) · Microelectronics Journal · doi:10.1016/j.mejo.2022.105569 · Cited by 14
Kushwaha, D., Sharma, A., Gupta, N., Raj, R., Joshi, A., Mishra, J., Kohli, R., Miryala, S., Joshi, R., Dasgupta, S., et al. (2022) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas48785.2022.9937908 · Cited by 15
“Biswas et al. [13] uses 10T bit-cell to address stability issues, but it occupies more area than other schemes.”
BackgroundRajput, A. K., & Pattanaik, M. (2023) · Microelectronics Journal · doi:10.1016/j.mejo.2023.105795 · Cited by 14
Wang, J., Zhang, T., Liu, S., Liu, Y., Wu, Y., Hu, S. G., Chen, T. P., Liu, Y., Yang, Y., & Huang, R. (2023) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2023.3304174 · Cited by 15
“Digital-to-analog converter (DAC) [15] or spikes’ encoder is required for the input, and ADC [2], [11], [16], [17], [18] or time-domain converter (TDC) [3], [19] is required 0018-9200 © 2023 IEEE.”
BackgroundHuang, P., Liu, X., Xin, Y., Gu, Y., Lee, A., Zhang, Y., Xu, Z., Chen, P., Zhang, Y. L., Deng, W., et al. (2024) · ACS Nano · doi:10.1021/acsnano.4c03278 · Cited by 15
An Integer-Floating-Point Dual-Mode Gain-Cell Computing-in-Memory Macro for Advanced AI Edge Chips
Wu, P.-C., Khwa, W.-S., Wu, J.-J., Su, J.-W., Jhang, C.-J., Chen, H.-Y., Ke, Z.-E., Chiu, T.-C., Hsu, J.-M., Cheng, C.-Y., et al. (2024) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2024.3470215 · Cited by 14
“C OMPUTING-IN-MEMORY (CIM) is a promising approach to overcoming the von Neumann bottle-neck [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19] by reducing the movement of data between system memory and computing devices.”
InfluentialDesign Framework for SRAM-Based Computing-In-Memory Edge CNN Accelerators
Wang, Y., Zou, Z., & Zheng, L. (2021) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas51556.2021.9401746 · Cited by 14
Charge Sharing and Charge Injection A/D Converters for Analog In-Memory Computing
Caselli, M., Papistas, I. A., Cosemans, S., Mallik, A., Debacker, P., & Verkest, D. (2021) · IEEE International New Circuits and Systems Conference (NEWCAS) · doi:10.1109/newcas50681.2021.9462775 · Cited by 14
“A different approach is proposed in [1], where a single column Fig.”
BackgroundMethodInfluentialDPCrypto: Acceleration of Post-Quantum Cryptography Using Dot-Product Instructions on GPUs
Lee, W., Seo, H., Hwang, S. O., Achar, R., Karmakar, A., & Mera, J. M. B. (2022) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2022.3176966 · Cited by 17
“Due to this reason, many research works have been devoted to design specific hardware for dot-product computation [16], [17], aimed towards providing a faster and more energy efficient implementations.”
BackgroundAn 8T SRAM Array with Configurable Word Lines for In-Memory Computing Operation
Zhang, J., Lin, Z., Wu, X., Peng, C., Lu, W., Zhao, Q., & Chen, J. (2021) · Electronics · doi:10.3390/electronics10030300 · Cited by 17
“Introduction Artificial intelligence and machine learning are highly valued technologies that are widely used in various fields [1].”
BackgroundQuantized convolutional neural networks: a hardware perspective
Zhang, L., Krestinskaya, O., Fouda, M. E., Eltawil, A. M., & Salama, K. N. (2025) · Frontiers in Electronics · doi:10.3389/felec.2025.1469802 · Cited by 18
The design of analogue in-memory computing tiles
Singh, A., Gallo, M. L., Vasilopoulos, A., Luquin, J., Narayanan, P., Burr, G. W., & Sebastian, A. (2025) · Nature Electronics · doi:10.1038/s41928-025-01537-5 · Cited by 21
“The encoding schemes may remain similar; however, the MVM read operation can be based on voltage 45,64,83–85 , time 86 or even charge 47,66,87–89 , as in the case of capacitor-based unit cells.”
Zhou, K., Jia, X., Zhao, C., Zhang, X., Wu, G., Mu, C., Zhu, H., Ding, Y., Chen, C., Xue, X., et al. (2022) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2022.3196678 · Cited by 13
“CIM can combine data access and computing operations by embedding computing functions into the memory, which can reduce the overhead of data movement, and realize large-scale parallel operation [6],[7].”
BackgroundCIM-MLC: A Multi-level Compilation Stack for Computing-In-Memory Accelerators
Qu, S., Zhao, S., Li, B., He, Y., Cai, X., Zhang, L., & Wang, Y. (2024) · International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) · doi:10.1145/3620665.3640359 · Cited by 18
“We sort out the designs of recent CIM accelerators from three dimensions: memory device, architecture hierarchy, and programming interface, and summarize them in Figure 1 [4, 6, 13, 18, 19, 21, 23, 28, 29, 33, 34, 39, 43, 46–51].”
BackgroundMethodInfluentialProcessing-in-Memory Technology for Machine Learning: From Basic to ASIC
Taylor, B., Zheng, Q., Li, Z., Li, S., & Chen, Y. (2022) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2022.3168404 · Cited by 14
Rhe, J., Moon, S., & Ko, J. H. (2022) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2022.3173649 · Cited by 15
A 1.23-GHz 16-kb Programmable and Generic Processing-in-SRAM Accelerator in 65nm
Sridharan, A., Angizi, S., Cherupally, S. K., Zhang, F., Seo, J.-S., & Fan, D. (2022) · European Solid-State Circuits Conference (ESSCIRC) · doi:10.1109/esscirc55480.2022.9911440 · Cited by 13
“BWNN Accelerators Generic Accelerators Reference PSRAM JSSC’19 [1] JSSC’19 [3] JSSC’20 [2] JSSC’18 [4] Technology 65nm 65nm 65nm 28nm 40nm Bit cell Density 8T 10T 8T 8T Transposable 10T Supply Voltage 0.”
BackgroundInfluentialRajput, A. K., Tiwari, A. K., & Pattanaik, M. (2023) · Circuits Systems and Signal Processing · doi:10.1007/s00034-022-02284-0 · Cited by 16
Lee, J., Kim, J.-H., Jo, W., Kim, S., Kim, S., & Yoo, H. (2021) · IEEE Micro · doi:10.1109/mm.2021.3096236 · Cited by 13
Xiao, K., Cui, X., Qiao, X., Wang, X., & Wang, Y. (2022) · Microelectronics Journal · doi:10.1016/j.mejo.2022.105506 · Cited by 12
Zhang, H., Jiang, L., Wu, J., Chen, T., Liu, J., Kang, W., & Zhao, W. S. (2022) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/3489517.3530398 · Cited by 12
“To perform SRAM-CIM in the analog domain for high energy-efficiency, a rich variety of designs have been reported, including currentdomain computing [14-19] and charge-domain computing [4], [20-24].”
MethodLi, Y., Du, L., & Du, Y. (2023) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas57966.2023.10168604 · Cited by 14
Dazzi, M., Sebastian, A., Parnell, T. P., Francese, P. A., Benini, L., & Eleftheriou, E. S. (2021) · IEEE Transactions on Computers · doi:10.1109/tc.2021.3073255 · Cited by 15
“There are several recent demonstrations of IMC using SRAM [16], [17], [18], [19], NOR Flash [20] as well as resistive memory devices [21], [22], [23].”
MethodIn-Memory Multibit Multiplication Based on Bitline Shifting
Zhang, J., Lin, Z., Wu, X., Tong, Z., Peng, C., Lu, W., Zhao, Q., Wu, H., & Chen, J. (2021) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2021.3099798 · Cited by 13
Chen, Z., Jin, Q., Wang, J., Wang, Y., & Yang, K. (2021) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1109/islped52811.2021.9502505 · Cited by 12
“The analog multiplication is performed in local capacitors in the cells and the accumulation is achieved by charge sharing [8], [11] or capacitive coupling [9], where no transistor nonideality is involved throughout the operations.”
BackgroundInfluentialZhang, H., Liu, J., Kang, W., Fan, Y., Fu, S., Bai, J., Pan, B., Liu, Y., & Zhao, W. S. (2021) · IEEE International Conference on Integrated Circuits, Technologies and Applications (ICTA) · doi:10.1109/icta53157.2021.9661898 · Cited by 11
8T-SRAM Based Process-In-Memory (PIM) System With Current Mirror for Accurate MAC Operation
Chung, K.-Y., Kim, H., An, Y., Seong, K., Shin, D.-H., Baek, K., & Shim, Y. S. (2024) · IEEE Access · doi:10.1109/access.2024.3358451 · Cited by 12
“To solve this issue, several techniques are actively researched [5], [6], [7], [9], [10], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], where some of the main ideas with 4-bit resolution are illustrated in Fig.”
MethodSACHI: A Stationarity-Aware, All-Digital, Near-Memory, Ising Architecture
Raman, S. R. S., John, L. K., & Kulkarni, J. P. (2024) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca57654.2024.00061 · Cited by 22
“With performance scaling, as (technology node) as mentioned in [2], the cycle time of SACHI can match modern CPUs.”
BackgroundMao, D., & Hao, Z. (2019) · Symmetry · doi:10.3390/sym11050703 · Cited by 11
“Start with LeNet-5 [24,25], a CNN particularly possesses normative structure-stacked convolutional layers (randomly followed by contrast max-pooling and normalization) followed by at least a completely linked layer.”
BackgroundUm, S., Kim, S., Kim, S., & Yoo, H. (2021) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2021.3067327 · Cited by 11
Prabhu, N. L., & Raghavan, N. (2021) · IEEE Access · doi:10.1109/access.2021.3136193 · Cited by 11
“[ 5 ].”
Sudarshan, C., Soliman, T., Lappas, J., Weis, C., Sadi, M. H., Jung, M., Guntoro, A., & Wehn, N. (2022) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2022.3170235 · Cited by 11
A Novel 8T XNOR-SRAM: Computing-in-Memory Design for Binary/Ternary Deep Neural Networks
Alnatsheh, N., Kim, Y., Cho, J., & Choi, K. K. (2023) · Electronics · doi:10.3390/electronics12040877 · Cited by 11
Memory-Immersed Collaborative Digitization for Area-Efficient Compute-in-Memory Deep Learning
Nasrin, S., Hashem, M. B., Darabi, N., Parpillon, B., Fahim, F., Gomes, W., & Trivedi, A. R. (2023) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas57966.2023.10168632 · Cited by 11
“T compute-in-SRAM/eDRAM [4], [14]–[16] as well as non-volatile memory crossbars [17], [18].”
MethodHsu, Y., Yao, C.-Y., Wu, T., Chiueh, T., & Liu, T.-T. (2021) · IEEE Solid-State Circuits Letters · doi:10.1109/lssc.2021.3103759 · Cited by 11
A Low-Power Analog Processor-in-Memory-Based Convolutional Neural Network for Biosensor Applications
Byun, S.-J., Kim, D., Park, K., Choi, Y.-J., Kumar, P., Ali, İ., Kim, D., Yoo, J.-M., Huh, H., Jung, Y., et al. (2022) · Sensors · doi:10.3390/s22124555 · Cited by 12
“Our design applied analog-based PIM to operate MAC filtering in RTL-based CNN architecture and CNN operations [17], PIM stores weight, advances MAC operations in parallel, and enables fast calculation.”
MethodSpintronic Materials and Devices for Artificial Intelligence Applications
Xu, M., Yue, X. Y., Jian, J., Li, T., Chai, Z., & Min, T. (2026) · Advanced Materials · doi:10.1002/adma.202518570 · Cited by 14
“Conventional memories such as SRAM [346, 347], dynamic random access memory (DRAM) [348, 349], and Flash [350, 351] have been explored for IMC but suffer from volatility, refresh overhead, or poor endurance.”
A Fully Differential 4-Bit Analog Compute-In-Memory Architecture for Inference Application
Kushwaha, D., Kohli, R., Mishra, J., Joshi, R., Dasgupta, S., & Bulusu, A. (2023) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas57966.2023.10168599 · Cited by 10
“Previous CIM works [7], [19], and [20] lack generating differential voltage changes on bitlines during multiplication, as shown in Fig.”
BackgroundMethodInfluentialBehavioral Level Simulation Framework to Support Error-Aware CNN Training with In-Memory Computing
Chang, S.-H., Liu, C. J., & Küster, A. (2022) · International Conference on Synthesis, Modeling, Analysis and Simulation Methods and Applications to Circuit Design · doi:10.1109/smacd55068.2022.9816307 · Cited by 10
Configurable Memory With a Multilevel Shared Structure Enabling In-Memory Computing
Zhao, Y., Lin, Z., Wu, X., Zhao, Q., Lu, W., Peng, C., Tong, Z., & Chen, J. (2022) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2022.3148327 · Cited by 10
Caselli, M., Debacker, P., & Boni, A. (2022) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2022.3174622 · Cited by 10
Tiny ci-SAR A/D Converter for Deep Neural Networks in Analog in-Memory Computation
Caselli, M., Bhattacharjee, D., Mallik, A., Debacker, P., & Verkest, D. (2022) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas48785.2022.9937613 · Cited by 10
“In the CS serial ADC in [10], fS is data dependent and it requires the synchronization of all the ADCs in the array, with a huge impact on the MVM computing speed when one of the column MAC results in the array is far from the expected value.”
BackgroundMethodDazzi, M., Sebastian, A., Francese, P. A., Parnell, T., Benini, L., & Eleftheriou, E. (2019) · arXiv · doi:10.48550/arxiv.1906.03474 · Cited by 10
“in-memory computing using SRAM [16, 9, 1, 15], NOR Flash [12] as well as resistive memory devices [18, 6, 14].”
MethodZhou, C., Garćıa-Redondo, F., Büchel, J., Boybat, I., Comas, X. T. I., Nandakumar, S. R., Das, S., Sebastian, A., Gallo, M. L., & Whatmough, P. N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.06503 · Cited by 10
“Many previous works demonstrate CiM using SRAM (Valavi et al., 2019; Biswas & Chandrakasan, 2018; Gonugondla et al., 2018; Jia et al., 2021; Lee et al., 2021; Sinangil et al., 2021).”
Yi, D., Lee, S., Choi, I., Yun, G., Choi, E. J., Park, J.-H., Kwak, J., Jang, S., Ha, S., Chang, I., et al. (2024) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2024.3372392 · Cited by 10
“While initial AMS-CIM systems supported low-precision weights such as binary and ternary weights [1], [2], [3], [4], recent state-of-the-art (SOTA) works have explored higher precision weights (e.g., 4-bit or 8-bit) to enhance DNN classification accuracy.”
BackgroundChen, T., Wang, W., Chen, J., Fu, H., Yi, W., Cheng, B., Zhang, H., & Pan, B. (2024) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2024.3384271 · Cited by 10
“A. Multi-Core CIM and CNN Accelerator CIM is a promising paradigm for solving the Von Neumann problem owing to its fused computing and memory architecture [16], [17], [18], [19], [20].”
BackgroundIn-memory computing: characteristics, spintronics, and neural network applications insights
Jangra, P., & Duhan, M. (2024) · Multiscale and Multidisciplinary Modeling Experiments and Design · doi:10.1007/s41939-024-00517-0 · Cited by 9
“Another advancement in the industry involves 10 T-SRAM cells (Biswas and Chandrakasan 2019), which maintain dual read paths without encountering read failures but at the cost of increased chip size due to additional transistors.”
Fang, Z., Wang, Y., & Xu, Y. (2025) · Microelectronics Journal · doi:10.1016/j.mejo.2025.106903 · Cited by 11
Dynamic Quantization Range Control for Analog-in-Memory Neural Networks Acceleration
Laubeuf, N., Doevenspeck, J., Papistas, I. A., Caselli, M., Cosemans, S., Vrancx, P., Bhattacharjee, D., Mallik, A., Debacker, P., Verkest, D., et al. (2022) · ACM Transactions on Design Automation of Electronic Systems · doi:10.1145/3498328 · Cited by 11
“Ad hoc methods to determine ADC quantization ranges have proven efficient on relatively small networks [7, 17, 24, 36].”
MethodIn-Memory Transposable Multibit Multiplication Based on Diagonal Symmetry Weight Block
Tong, Z., Zhao, Y., Zhang, J., Lin, Z., Lin, X., & Wu, X. (2023) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2023.3266597 · Cited by 9
“Considering the overhead of quantization circuits, most existing CIM works have selected multiplexing quantization circuits [10], [11], [12]; however, these efforts reduce the throughput of the operations.”
DRAM Translation Layer: Software-Transparent DRAM Power Savings for Disaggregated Memory
Jin, W., Jang, W., Park, H., Lee, J.-S., Kim, S., & Lee, J. W. (2023) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3579371.3589051 · Cited by 10
“The power and area estimation can be normalized to 25.7mW and 0.165 𝑚𝑚 2 at 7nm technology, assuming both are proportional to the ( 𝑇𝑒𝑐ℎ𝑛𝑜𝑙𝑜𝑔𝑦 ) 2 following the methodology of Biswas and Chan-drakasan [11].”
BackgroundInfluentialAn Area-Efficient In-Memory Implementation Method of Arbitrary Boolean Function Based on SRAM Array
Zhang, S., Cui, X., Wei, F., & Cui, X. (2023) · IEEE Transactions on Computers · doi:10.1109/tc.2023.3301156 · Cited by 11
“row[3] RW-cells(Region1) row[4] row[5] row[6] row[7] RW-cells (Region2) row[8] row[9] row[10] row[11] RW-cells(Region3) row[12] row[13] row[14] row[15] Fig.”
BackgroundA 9T-SRAM in-memory computing macro for Boolean logic and multiply-and-accumulate operations
Dai, C., Ren, Z., Guan, L., Liu, H., Gao, M., Lu, W., Pang, Z., Peng, C., & Wu, X. (2023) · Microelectronics Journal · doi:10.1016/j.mejo.2023.106087 · Cited by 9
Sadasivuni, S., Damodaran, V., Banerjee, I., & Sanyal, A. (2022) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas54282.2022.9869980 · Cited by 9
Ramanathan, A. K., Rangachar, S. S., Govindarajan, H. T., Hung, J.-M., Lee, C., Xue, C.-X., Huang, S.-P., Hsueh, F.-K., Shen, C., Shieh, J., et al. (2021) · IEEE Journal on Exploratory Solid-State Computational Devices and Circuits · doi:10.1109/jxcdc.2021.3087745 · Cited by 9
Maharmeh, H. A., Sarhan, N. J., Ismail, M., & Alhawari, M. (2023) · IEEE Open Journal of Circuits and Systems · doi:10.1109/ojcas.2023.3332853 · Cited by 9
“…challenges in digital systems, analog and mixed-signal hardware accelerators have been explored to build artificial neural networks (ANNs) that can outperform the digital-based ones by several orders of magnitudes in energy efficiency, computation, and training time [22], [23], [24], [25], [26].”
BackgroundMethodA 28 nm Dual-Mode SRAM-CIM Macro With Local Computing Cell for CNNs and Grayscale Edge Detection
Peng, C., Chen, X., Gao, M., Guo, J., Guan, L., Dai, C., Lin, Z., & Wu, X. (2025) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2025.3578319 · Cited by 9
“Since the 6T array structure was proposed [3], [11], [12], [13], [14], [15], researchers have expanded this design to include 8T, Twin 8T, and 10T bitsells, [16], [17], [18], [19].”
Lee, S., & Kim, Y. (2024) · Electronics · doi:10.3390/electronics13030666 · Cited by 9
Li, X., Gao, M., Ren, Z., Yu, K., Lu, W., Dai, C., Hu, W. S., Peng, C., & Wu, X. (2024) · Microelectronics Journal · doi:10.1016/j.mejo.2024.106124 · Cited by 8
Liu, Y., Zhao, R., Xiao, H., Liu, Y., & Chen, J. (2024) · AEU - International Journal of Electronics and Communications · doi:10.1016/j.aeue.2024.155346 · Cited by 8
Reconfigurable Precision SRAM-based Analog In-memory-compute Macro Design
Bazzi, J., Jamil, R., ElHajj, D., Kanj, R. N., Fouda, M. E., & Eltawil, A. M. (2024) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas58744.2024.10558662 · Cited by 8
AFSRAM-CIM: Adder Free SRAM-Based Digital Computation-in-Memory for BNN
Arrassi, A. E., Yaldagard, M. A., Tao, X., Shahroodi, T., Mir, F. J., Biyani, Y., Gomony, M. D., Gebregiorgis, A., Joshi, R., & Hamdioui, S. (2024) · IFIP/IEEE International Conference on Very Large Scale Integration · doi:10.1109/vlsi-soc62099.2024.10767789 · Cited by 8
“The work in [25] proposes a 10-”
Mu, J., & Kim, B. (2020) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas45731.2020.9181104 · Cited by 8
Zeng, S., Zhang, Y., Zhu, Z., Qin, Z., Dou, C., Si, X., & Li, Q. (2021) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas51828.2021.9458438 · Cited by 8
“To obtain a better energy efficiency, previous CIM researches mainly focus on the acceleration of low-bit-precision neural networks, such as binary weight neural networks (BWN) [3]-[5],[9].”
MethodCompute-in-Memory Upside Down: A Learning Operator Co-Design Perspective for Scalability
Nasrin, S., Shukla, P., Jaisimha, S., & Trivedi, A. R. (2021) · Design, Automation, and Test in Europe · doi:10.23919/date51398.2021.9474119 · Cited by 9
Kim, J., & Park, J. (2021) · International SoC Design Conference · doi:10.1109/isocc53507.2021.9613938 · Cited by 8
“The current domain IMC can operate multi-bit Input Activation (iAct.) by using Digital-Analog Converter (DAC), but suffers from Non-linearity in MAC operation such as in [1] and [2].”
MethodInfluentialBinary is All You Need: Ultra-Efficient Arrhythmia Detection with a Binary-Only Compressive System
Tian, F., Wang, X., Chen, J., Yang, J., Sawan, M., Tsui, C.-Y., & Cheng, K.-T. T. (2023) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas57966.2023.10168576 · Cited by 8
“One promising approach to implement CIM is using SRAM (SRAM-CIM) [15].”
MethodHarada, M., Takahashi, M., Sakai, S., & Morie, T. (2020) · Japanese Journal of Applied Physics · doi:10.35848/1347-4065/ab7ff1 · Cited by 10
Energy Efficient, Hamming Code Technique for Error Detection/Correction Using In-Memory Computation
Mythrai, Pragna, Kavitha, S., Singh, P., Shah, A. P., Vishwakarma, S., & Reniwal, B. S. (2021) · International Symposium on VLSI Design and Test (VDAT) · doi:10.1109/vdat53777.2021.9601068 · Cited by 8
Kim, H., An, Y., Kim, R., Kim, S. Y., & Shim, Y. S. (2024) · IEEE Access · doi:10.1109/access.2024.3487975 · Cited by 7
“1(a), several types of SRAM based CIM structure has been investigated, namely Analog CIM and Digital CIM, where the Analog CIM is further categorized into two groups: current domain [3], [4], [5], [6], [7], [8], [9], [10] and charge domain [11], [12], [13], [14], [15], [16], [17].”
MethodZhou, Y., Yi, W., Sun, S., Wang, W., Bai, J., Zhang, H., & Kang, W. (2025) · Asia and South Pacific Design Automation Conference (ASP-DAC) · doi:10.1145/3658617.3697580 · Cited by 7
Area-Optimized 2D Interleaved Adder Tree Design for Sparse DCIM Edge Processing
Sankhe, A., Lokhande, M., Sharma, R., & Vishvakarma, S. K. (2025) · International Symposium on Quality Electronic Design (ISQED) · doi:10.1109/isqed65160.2025.11014431 · Cited by 10
“In the last decade, several researchers have improved DCIM-based Edge-AI accelerators at algorithmic codesign [9], architecture [5], [7], [8], [12], and circuit level [4], [5], [13], [14], enabling opportunities to address challenges related to computational energy, memory bandwidth, SNR and…”
Measurement-driven neural-network training for integrated magnetic tunnel junction arrays
Borders, W. A., Madhavan, A., Daniels, M. W., Georgiou, V., Lueker-Boden, M., Santos, T., Braganca, P. M., Stiles, M. D., McClelland, J. J., & Hoskins, B. D. (2024) · Physical Review Applied · doi:10.1103/physrevapplied.21.054028 · Cited by 13
HfO2-Based Synaptic Spiking Neural Network Evaluation to Optimize Design and Testing Cost
Tushar, S., Das, H., & Rose, G. S. (2024) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas58744.2024.10558518 · Cited by 7
“Therefore, in memory computing system has been developed recently to overcome the von Neuman bottleneck [10], [11], [12].”
BackgroundWang, Y., Zhou, Y., Wang, B., Xiong, T., Kong, Y., & Si, X. (2021) · International Conference on UK-China Emerging Technologies (UCET) · doi:10.1109/ucet54125.2021.9674995 · Cited by 7
Boosting RRAM-Based Mixed-Signal Accelerators in FD-SOI Technology for ML Applications
Boni, A., Malena, F., Saccani, F., Amoretti, M., & Caselli, M. (2023) · IEEE Journal on Exploratory Solid-State Computational Devices and Circuits · doi:10.1109/jxcdc.2023.3309713 · Cited by 7
“Accelerators exploit the concept of analog in-memory computation (AiMC) to reduce data movement between memory and processor [3].”
BackgroundAn SRAM-Based Reconfigurable Cognitive Computation Matrix for Sensor Edge Applications
Peng, S.-Y., Liu, I.-C., Wu, Y.-H., Lin, T.-J., Chen, C.-J., Li, X.-Z., Cheng, Y., Lin, P.-H., Hung, K.-H., & Tsao, Y. (2023) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2023.3303910 · Cited by 8
“Binary and ternary neural networks with single-bit weights were first developed for simple networks [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27].”
BackgroundMethodInfluentialXiao, S., Li, F., Hao, T., Xiao, L., Xiao, M., Mao, W., & Han, G. (2025) · Integrated Circuits and Systems · doi:10.23919/ics.2025.3571019 · Cited by 7
Time-Multiplexed Flash ADC for Deep Neural Network Analog in-Memory Computing
Boni, A., Frattini, F., & Caselli, M. (2021) · IEEE International Conference on Electronics, Circuits and Systems (ICECS) · doi:10.1109/icecs53924.2021.9665494 · Cited by 6
BDAPbI4 Dion Jacobson hybrid perovskite-based artificial nociceptors on biodegradable substrate
Khemnani, M., Thakkar, P., Lokhandwala, A., Andola, B., Tripathi, B., Srivastava, Y. K., Srivastava, A., & Solanki, A. (2024) · Sensors and Actuators A Physical · doi:10.1016/j.sna.2024.115382 · Cited by 6
A real-time and energy-efficient SRAM with mixed-signal in-memory computing near CMOS sensors
Diaz-Madrid, J.-A., Doménech‐Asensi, G., Ruíz, R., & Zapata, J. (2024) · Journal of Real-Time Image Processing · doi:10.1007/s11554-024-01520-x · Cited by 6
Intelligent Sensing Framework: Near-Sensor Machine Learning for Efficient Data Transmission
Huang, W. M., Rezvani, A., Chen, H., Ni, Y., Yun, S., Jeong, S., Zhang, G., & Imani, M. (2024) · IEEE Sensors Journal · doi:10.1109/jsen.2024.3440988 · Cited by 16
“Furthermore, some research have placed emphasis on the optimization of hardware structures, thereby enhancing the efficiency of edge computing [21], [22], [23].”
BackgroundChung, S., & Wang, J. (2019) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2019.2934929 · Cited by 7
Saragada, P. K., & Das, B. P. (2022) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2022.3199396 · Cited by 6
“The proposed TC-IMC architecture performs multibit MAC operation without using extra peripheral circuitry such as DAC array [1], [2], [17]”
BackgroundMethodInfluential
Metadata
- Authors
- Biswas, A., & Chandrakasan, A. P.
- Affiliations
- Texas Instruments (United States) (US)
- Massachusetts Institute of Technology (US)
- Venue
- IEEE Journal of Solid-State Circuits
- Published
- 2019-01
- Volume
- 54
- Issue
- 1
- Pages
- 217-230
- ISSN
- 0018-9200 (ISSN-L)
- 1558-173X
- OpenAlex
- W2904299207
- Cited by
- 414
- References
- 29
- Citations by year
- 2026: 36 · 2025: 57 · 2024: 75 · 2023: 75 · 2022: 74 · 2021: 59 · 2020: 24 · 2019: 13
- FWCI
- 9.17
- Type
- journal article
- Language
- en
- Open access
- green
- hdl.handle.net
- Licence
- Locations
- IEEE Journal of Solid-State Circuits (publishedVersion)
- DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc-sa)
- Funders
- Intel Corporation
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Low-power high-performance VLSI design
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Analog and Mixed-Signal Circuit Design
- Keywords
- SRAM, in-memory computing, dot product computation, energy efficiency, low-power neural network hardware, analog-to-digital converter, MNIST classification
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.