{"title":"A 14 nm MRAM-Based Multi-bit Analog In-Memory Computing With Process-Variation Calibration for 72 Macros-Based Accelerator","authors":"Sungmeen Myung;Seok-Ju Yun;Minje Kim;Wooseok Yi;Jaehyuk Lee;Jangho An;Kyoung-Rog Lee;Chang-Woo Shin;Seungchul Jung;Soonwan Kwon","doi":"10.1109/LSSC.2024.3465595","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3465595","url":null,"abstract":"This letter presents an analog in-memory computing (IMC) macro utilizing 14 nm MRAM technology. To facilitate energy-efficient high-throughput multiply accumulate (MAC) operations, a multi-bit weight is introduced using stacked magnetic tunnel junction architecture and an analog bit-parallel MAC (ABP-MAC) scheme is proposed. This approach delivers 3.3 times better TOPS/mm2 than the state-of-the-art MRAM-based IMC macro. Additionally, a comprehensive calibration technique significantly improves computational accuracy across 72 IMC macros. The proposed IMC macro achieves 18.29 TOPS/mm2 and 340.8 TOPS/W with 1-bit normalization and classification accuracy of 90.2% with the Google speech commands dataset.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"295-298"},"PeriodicalIF":2.2,"publicationDate":"2024-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Meng Wu;Wenjie Ren;Peiyu Chen;Wentao Zhao;Tianyu Jia;Le Ye
{"title":"S2D-CIM: SRAM-Based Systolic Digital Compute-in-Memory Framework With Domino Data Path Supporting Flexible Vector Operation and 2-D Weight Update","authors":"Meng Wu;Wenjie Ren;Peiyu Chen;Wentao Zhao;Tianyu Jia;Le Ye","doi":"10.1109/LSSC.2024.3463697","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3463697","url":null,"abstract":"In this letter, we propose an SRAM-based systolic digital compute-in-memory (S2D-CIM) framework which enables flexible input dataflow and mapping strategy to enhance the effective energy efficiency (EE), area efficiency, and writing bandwidth for practical CIM with innovations: 1) multistage domino data path (DDP); 2) a configurable asynchronous timing scheme; and 3) a 2-D burst writing scheme. The proposed S2D-CIM is fabricated using TSMC 22-nm technology and achieves 9.19 and 24.4 TOPS/W peak EE in systolic mode and broadcast mode, respectively, at full precision of 8-bit input, 8-bit weight, and 21-bit output. Compared with state of the arts, it achieves \u0000<inline-formula> <tex-math>$1.67times $ </tex-math></inline-formula>\u0000 effective EE improvement. Thanks to reusing introduced DDP, fast 2-D weight update is realized and gains 1.187 Tb/s writing bandwidth, which is \u0000<inline-formula> <tex-math>$14.3times $ </tex-math></inline-formula>\u0000 better than that of normal SRAM macro with the same capacity.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"291-294"},"PeriodicalIF":2.2,"publicationDate":"2024-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Reconfigurable Floating-Point Compute-in-Memory With Analog Exponent Preprocesses","authors":"Pengyu He;Yuanzhe Zhao;Heng Xie;Yang Wang;Shouyi Yin;Li Li;Yan Zhu;Rui P. Martins;Chi-Hang Chan;Minglei Zhang","doi":"10.1109/LSSC.2024.3463208","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3463208","url":null,"abstract":"This letter presents a reconfigurable floating-point compute-in-memory (FP-CIM) macro that preprocesses the exponent in the analog domain, enhancing the energy efficiency of edge devices for the floating-point (FP) inference. The presented FP-CIM macro supports FP8 inference, while can be configured to BP16 precision in a segmented computation manner. Furthermore, a time-domain analog-to-digital converter facilitates the analog compute-in-memory (CIM) macro while improving energy efficiency by sharing the counter and quantizing in a coarse-fine structure. Fabricated in a 28-nm CMOS process, the presented FP-CIM macro achieves 314.6-TFLOPS/W energy efficiency and 12.13-TFLOPS/mm2 area efficiency at the FP8 mode.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"271-274"},"PeriodicalIF":2.2,"publicationDate":"2024-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438608","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kim-Hoang Nguyen;Quyet Nguyen;Quynh-Trang Nguyen;Thanh-Tung Vu;Woojin Ahn;Loan Pham-Nguyen;Hanh-Phuc Le;Minkyu Je
{"title":"A Fully Integrated Dynamic-Voltage-Scaling Stimulator IC for Cochlear Implants","authors":"Kim-Hoang Nguyen;Quyet Nguyen;Quynh-Trang Nguyen;Thanh-Tung Vu;Woojin Ahn;Loan Pham-Nguyen;Hanh-Phuc Le;Minkyu Je","doi":"10.1109/LSSC.2024.3462559","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3462559","url":null,"abstract":"A fully integrated dynamic-voltage-scaling stimulator IC, consisting of a novel reconfigurable supply modulator (RSM) and 12 high-voltage-tolerant channel drivers, for cochlear implants, is presented, utilizing a 180-nm standard CMOS process. The RSM is designed to adaptively generate one of four supply voltage levels ranging from 2.6 to 11.3 V, effectively stimulating the cochlea with varying electrode-tissue-interface impedance and stimulus currents while offering improved power efficiency. The channel driver design is miniaturized to support high-channel-count applications within a single IC. Additional excessive current protection is implemented to ensure charge balancing between biphasic stimulating pulses, complementing the electrode-shorting technique.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"275-278"},"PeriodicalIF":2.2,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Dongha Lee;Seki Kim;Takahiro Nomiyama;Dong-Hoon Jung;Dongsu Kim;Jongwoo Lee
{"title":"A Computational Digital LDO With Distributed Power-Gating Switches and Time-Based Fast-Transient Controller for Mobile SoC Application","authors":"Dongha Lee;Seki Kim;Takahiro Nomiyama;Dong-Hoon Jung;Dongsu Kim;Jongwoo Lee","doi":"10.1109/LSSC.2024.3461158","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3461158","url":null,"abstract":"This letter introduces a 10 A computational digital LDO (CDLDO) for mobile SoC application specifically targeting a big CPU core. The proposed CDLDO eliminates the power-FET area overhead by reusing power gating switches (PGSs) already distributed throughout the entire CPU. The CDLDO employs a time-based exponential control (TEC) with a slope detector to achieve fast-transient response and improve stability. Furthermore, a step-back and a negative-step control are introduced to mitigate the effect of the propagation delay between the controller and the PGSs. Additionally, a pre-computational scheme significantly reduces calculation time and relaxes timing constraints during synthesis. The proposed CDLDO is implemented in 3 nm GAAFET CMOS process. An implemented IC of eight distributed CDLDO units provides a maximum load current of 10 A with a current density of 263 A/mm2. The CDLDO shows 94 mV droop under 6.5 A/1 ns load transition.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"287-290"},"PeriodicalIF":2.2,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438609","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yicheng Wang;Zhaowu Wang;Zhenyu Wang;Xiaochen Tang;Yong Wang
{"title":"An X-Band Expandable Reconfigurable 1:2 Power Divider Switch for Switched Beam-Forming Networks in 0.10-µm GaAs Process","authors":"Yicheng Wang;Zhaowu Wang;Zhenyu Wang;Xiaochen Tang;Yong Wang","doi":"10.1109/LSSC.2024.3458453","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3458453","url":null,"abstract":"In this letter, an X-band expandable reconfigurable 1:2 power divider switch (PDSW) is proposed for switched beam-forming networks. A switched inductor-artificial transmission line (SI-ATL) is proposed. With proper switch logic, the SI-ATL features two types of transmission line (TL): 1) a \u0000<inline-formula> <tex-math>$lambda $ </tex-math></inline-formula>\u0000/4 TL of \u0000<inline-formula> <tex-math>$50sqrt {2} ; Omega $ </tex-math></inline-formula>\u0000 and 2) a TL of \u0000<inline-formula> <tex-math>$50 ; Omega $ </tex-math></inline-formula>\u0000. This enables the PDSW to realize three port states of two corresponding modes, including single-pole–double-throw (SPDT) mode and power divider (PD) mode. The PDSW has the ability to be expanded to an N-stage 1:\u0000<inline-formula> <tex-math>$2^{N}$ </tex-math></inline-formula>\u0000 matrix with \u0000<inline-formula> <tex-math>$2^{2^{N}} - 1$ </tex-math></inline-formula>\u0000 states. The proposed design is fabricated with a 0.10-\u0000<inline-formula> <tex-math>$mu $ </tex-math></inline-formula>\u0000m GaAs pHEMT process. The measurement results show a \u0000<inline-formula> <tex-math>$leq 1$ </tex-math></inline-formula>\u0000.2-dB insertion loss (IL), a \u0000<inline-formula> <tex-math>$geq $ </tex-math></inline-formula>\u0000 10-dB return loss (RL), and a \u0000<inline-formula> <tex-math>$geq 40$ </tex-math></inline-formula>\u0000-dBm input 3rd-order intercept points (IIP3), in both modes. The isolation is 27–32 dB for SPDT mode and 14–31 dB for PD mode.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"303-306"},"PeriodicalIF":2.2,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430743","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A 112-Gb/s, -10 dBm Sensitivity, +5 dBm Overload, and SiPh-Based Receiver Frontend in 22-nm FDSOI","authors":"Mahdi Parvizi;Bahar Jalali;Toshi Omori;John Rogers;Li Chen;Long Chen;Ricardo Aroca","doi":"10.1109/LSSC.2024.3457775","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3457775","url":null,"abstract":"This letter demonstrates a Si-Photonic (SiPh)-based 112 Gb/s PAM4 optical receiver frontend using novel single-ended transimpedance amplifier (TIA) architecture that achieves −10 and +5 dBm input optical modulation amplitude (OMA) sensitivity and overload, respectively. To achieve that an overload mitigation circuit is proposed to break the tradeoff between noise and linearity of the shunt feedback CMOS TIAs. The TIA is optimized to provide the best sensitivity and linearity performance at minimum and maximum input OMA, respectively. Implemented in 22-nm FDSOI technology, and designed for 112 Gb/s PAM4 optical links, the TIA achieves more than +15 dBm OMA range with 11 pA/\u0000<inline-formula> <tex-math>$surd $ </tex-math></inline-formula>\u0000Hz input referred noise while burning only 155 mW from an 1.8-V supply.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"263-266"},"PeriodicalIF":2.2,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142276431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Fully Integrated 5510-μm² Process Monitor and Threshold Voltage Extractor Circuit in 28 nm","authors":"Ido Shpernat;Asaf Feldman;Joseph Shor","doi":"10.1109/LSSC.2024.3457768","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3457768","url":null,"abstract":"A new architecture of an on-die process monitor circuit is demonstrated in 28 nm. The proposed circuit can extract the threshold voltage, \u0000<inline-formula> <tex-math>$V_{mathrm { TH,}}$ </tex-math></inline-formula>\u0000 and random mismatch of a transistor using multiple extraction methods, including the second derivative method. A sigma-delta modulator analog-to-digital converter samples the output to enable on-die processing of the results. A \u0000<inline-formula> <tex-math>$V_{mathrm { DS}}$ </tex-math></inline-formula>\u0000 voltage control loop enables \u0000<inline-formula> <tex-math>$V_{mathrm { TH}}$ </tex-math></inline-formula>\u0000 extraction in both the linear and saturation regions of the device. The circuit has a compact area of \u0000<inline-formula> <tex-math>$5510~mu $ </tex-math></inline-formula>\u0000m2.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"279-282"},"PeriodicalIF":2.2,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438622","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Design Challenges of Fully Integrated DC–DC Converters for Modern Power Delivery Architectures","authors":"Suyang Song;Alessandro Novello;Taekwang Jang","doi":"10.1109/LSSC.2024.3457272","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3457272","url":null,"abstract":"This letter presents recent design challenges of modern power delivery architectures and circuit techniques for them. Recent computational loads impose significant power output demands on dc-dc converters while ever-shrinking Internet of Things (IoT) systems demand dc-dc converters with small footprints. Consequently, fully integrated dc-dc converters are highly desirable in contemporary power delivery architectures thanks to their compact footprint, high power density, and fast output regulation. However, numerous challenges exist in fully integrating dc-dc converters, necessitating the investigation of various circuit topologies and complex regulation schemes to ensure proper operation and versatility.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"267-270"},"PeriodicalIF":2.2,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142328412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A D-Band 13-mW Dual-Mode CMOS LNA for Joint Radar–Communication in 22-nm FD-SOI CMOS","authors":"Shankkar Balasubramanian;Kristof Vaesen;Anirudh Kankuppe;Sehoon Park;Carsten Wulff","doi":"10.1109/LSSC.2024.3455889","DOIUrl":"https://doi.org/10.1109/LSSC.2024.3455889","url":null,"abstract":"This letter presents a D-band low-noise amplifier (LNA) for joint radar-communication applications in 22-nm CMOS technology. The 4-stage LNA uses transistor switching and bias class changes to achieve dual-mode functionality. In the radar mode, the LNA achieves gain of 17 dB, noise figure (NF) of 7.7 dB, 3-dB bandwidth (BW) of 117–129 GHz, and IP1dB of −20 dBm, respectively. In the communication mode, the LNA achieves gain of 22.6 dB, NF of 8.5 dB, BW of 115.9–128.9 GHz, and IP1dB of −29 dBm, respectively. The power consumption for the radar and communication modes is 13 and 12.2 mW, respectively. The LNA has a core area of \u0000<inline-formula> <tex-math>$0.06~text {mm}^{2}$ </tex-math></inline-formula>\u0000.","PeriodicalId":13032,"journal":{"name":"IEEE Solid-State Circuits Letters","volume":"7 ","pages":"259-262"},"PeriodicalIF":2.2,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10669787","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142276432","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}