Chih-Cheng Hsieh
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Chih-Cheng Hsieh has a total of 184 co-authors. Here are the top co-authors by number of shared publications:
- Kea-Tiong Tang
National Tsing Hua University
27 shared publications - Meng-Fan Chang
National Tsing Hua University
26 shared publications - Ren-Shuo Liu
Department of Electrical Engineering, National Tsing Hua University (NTHU), Hsinchu, Taiwan
16 shared publications Chung-Chuan Lo
Institute of Systems Neuroscience, National Tsing Hua University (NTHU), Hsinchu, Taiwan
14 shared publicationsRen-Shuo Liu
Department of Electrical Engineering, National Tsing Hua University (NTHU), Hsinchu, Taiwan
9 shared publications- Chung-Chuan Lo
National Tsing Hua University
8 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 73  (11) , 7944-7954
A 128 × 128 Dual-Resolution CMOS Image Sensor With In-Sensor Temporal and Spatial Gradient Calculations for Optical Flow Pre-Processing
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 73  (11) , 7944-7954
A 128 × 128 Dual-Resolution CMOS Image Sensor With In-Sensor Temporal and Spatial Gradient Calculations for Optical Flow Pre-Processing
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34  (2) , 332-342
A 15-bit ROIC With Current-Mode Folding Integration Technique for Microbolometer Thermal Imagers
This article presents a 52-mK noise-equivalent temperature difference (NETD) reference-cell-free wide temperature sensing range readout integrated circuit (ROIC) for a microbolometer thermal imager. The proposed ROIC utilizes a 15-bit hybrid analog-to-digital converter (ADC) that employs the current-mode folding integration (CMFI) technique for the coarse conversion and single-slope (SS) operation for the fine conversion. With the proposed coarse CMFI operation, the ROIC achieves a$300\times $ larger conversion range and$300\times $ higher conversion gain compared to the conventional nonfolding integration (NFI) technique. Moreover, due to the extended conversion range, the ROIC is able to cover the sensor’s PVT variations without any signal saturation. With the hybrid operation of CMFI and SS, the ROIC achieves a differential nonlinearity (DNL) and an integral nonlinearity (INL) of$\pm 0.04~^{\circ }$ C and$\pm 0.12~^{\circ }$ C, respectively. Furthermore, it achieves a wide temperature sensing range from$- 20~^{\circ }$ C to$107~^{\circ }$ C while considering ±15% chip-to-chip and ±3% cell-to-cell sensor variations.- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34  (2) , 332-342
A 15-bit ROIC With Current-Mode Folding Integration Technique for Microbolometer Thermal Imagers
This article presents a 52-mK noise-equivalent temperature difference (NETD) reference-cell-free wide temperature sensing range readout integrated circuit (ROIC) for a microbolometer thermal imager. The proposed ROIC utilizes a 15-bit hybrid analog-to-digital converter (ADC) that employs the current-mode folding integration (CMFI) technique for the coarse conversion and single-slope (SS) operation for the fine conversion. With the proposed coarse CMFI operation, the ROIC achieves a$300\times $ larger conversion range and$300\times $ higher conversion gain compared to the conventional nonfolding integration (NFI) technique. Moreover, due to the extended conversion range, the ROIC is able to cover the sensor’s PVT variations without any signal saturation. With the hybrid operation of CMFI and SS, the ROIC achieves a differential nonlinearity (DNL) and an integral nonlinearity (INL) of$\pm 0.04~^{\circ }$ C and$\pm 0.12~^{\circ }$ C, respectively. Furthermore, it achieves a wide temperature sensing range from$- 20~^{\circ }$ C to$107~^{\circ }$ C while considering ±15% chip-to-chip and ±3% cell-to-cell sensor variations.- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 61  (1) , 225-236
A μ-NMC-Δ-IMC Heterogeneous STT-MRAM Compute-in-Memory Macro Using Δ-Clamping Bit Reduction for Noise-Tolerant Bayesian Neural Networks
Nonvolatile compute-in-memory (nvCIM) macros integrate memory and computation to enable high-density multiply-and-accumulate (MAC) acceleration for inference tasks on low-power edge artificial intelligence (AI) devices. Bayesian neural networks (BNNs)—which represent weights with mean ($\mu $ ) and variance-derived deviation ($\Delta $ )—have emerged as a promising inference model to enhance robustness against noisy real-world inputs. However, deploying BNNs on nvCIM platforms poses several design challenges: 1) ensuring reliable in situ$\Delta $ -weight generation using stochastic memory devices under write-monitor sense margin (WM-SM) degradation; 2) balancing inference accuracy against energy efficiency in dual-weight MAC operations; 3) preventing in-memory-compute (IMC) MAC sense margin (SM) degradation in large memory arrays; and 4) avoiding excessive MAC energy consumption due to bit-level$\Delta $ -weight redundancy. This work addresses these challenges using a heterogeneous nvCIM macro featuring four key schemes: 1) self-compare write-termination (SCWT); 2) a heterogeneous nvCIM structure using near-memory compute (NMC) for$\mu $ -compute and IMC for$\Delta $ -compute ($\mu {N}$ -$\Delta {I}$ ); 3) 2-D clamping-voltage scaling with sense margin compensation (2D-CVS-SMC); and 4)$\Delta $ -clamping bit reduction (DCBR). A 4.125-Mb macro fabricated using foundry-provided 22-nm spin-transfer-torque magnetic random access memory (STT-MRAM) supports 8-bit MAC operations with 18-ns MAC computing latency, 120.8-TOPS/W energy efficiency, and noise-tolerant BNN inference—demonstrating its effectiveness for next-generation edge AI applications.- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 61  (1) , 225-236
A μ-NMC-Δ-IMC Heterogeneous STT-MRAM Compute-in-Memory Macro Using Δ-Clamping Bit Reduction for Noise-Tolerant Bayesian Neural Networks
Nonvolatile compute-in-memory (nvCIM) macros integrate memory and computation to enable high-density multiply-and-accumulate (MAC) acceleration for inference tasks on low-power edge artificial intelligence (AI) devices. Bayesian neural networks (BNNs)—which represent weights with mean ($\mu $ ) and variance-derived deviation ($\Delta $ )—have emerged as a promising inference model to enhance robustness against noisy real-world inputs. However, deploying BNNs on nvCIM platforms poses several design challenges: 1) ensuring reliable in situ$\Delta $ -weight generation using stochastic memory devices under write-monitor sense margin (WM-SM) degradation; 2) balancing inference accuracy against energy efficiency in dual-weight MAC operations; 3) preventing in-memory-compute (IMC) MAC sense margin (SM) degradation in large memory arrays; and 4) avoiding excessive MAC energy consumption due to bit-level$\Delta $ -weight redundancy. This work addresses these challenges using a heterogeneous nvCIM macro featuring four key schemes: 1) self-compare write-termination (SCWT); 2) a heterogeneous nvCIM structure using near-memory compute (NMC) for$\mu $ -compute and IMC for$\Delta $ -compute ($\mu {N}$ -$\Delta {I}$ ); 3) 2-D clamping-voltage scaling with sense margin compensation (2D-CVS-SMC); and 4)$\Delta $ -clamping bit reduction (DCBR). A 4.125-Mb macro fabricated using foundry-provided 22-nm spin-transfer-torque magnetic random access memory (STT-MRAM) supports 8-bit MAC operations with 18-ns MAC computing latency, 120.8-TOPS/W energy efficiency, and noise-tolerant BNN inference—demonstrating its effectiveness for next-generation edge AI applications.- 15 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 61  (1) , 211-224
A Microscaling Multi-Mode Gain-Cell Computing-in-Memory Macro for Advanced AI Edge Device
The microscaling (MX) format is an emerging data representation that quantizes high-bitwidth floating-point (FP) values into low-bitwidth FP-like values with a shared-scale (SS) exponent. When implemented with computing-in-memory (CIM), MX allows an attractive tradeoff between accuracy and hardware efficiency for specific neural network (NN) workloads. This work presents the first multi-mode gain-cell (GC) CIM macro capable of processing MX, integer (INT), and FP multiply-and-accumulate (MAC) operations with high energy efficiency (EEF) and area efficiency (AEF). The proposed macro employs four important innovations: 1) a multi-mode input processing unit (M2-IPU) with SS-variance-aware MAC flow (SS-VAF) for SS processing and SS alignment within the CIM macro to reduce system-to-CIM data transfer and compute energy; 2) a pattern-aware hybrid adder tree (PAH-ADT), which improves EEF and AEF by optimizing the common input patterns; 3) an accumulation-aware data flow (A2-DF) that adjusts the write path based on accumulation size to reduce data transfer energy; and 4) a 3.xT GC, which boosts data retention time (DRT) by increasing parasitic capacitance without additional area overhead. A 16-nm FinFET 216-kb MX–INT–FP multi-mode GC-CIM macro achieved 133.5 TFLOPS/W for MX-MAC with MXINT8 input, MXINT8 weight, and FP32 output; and 91.9 TFLOPS/W for FP-MAC with BF16 input, BF16 weight, and FP32 output.
- 15 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 61  (1) , 211-224
A Microscaling Multi-Mode Gain-Cell Computing-in-Memory Macro for Advanced AI Edge Device
The microscaling (MX) format is an emerging data representation that quantizes high-bitwidth floating-point (FP) values into low-bitwidth FP-like values with a shared-scale (SS) exponent. When implemented with computing-in-memory (CIM), MX allows an attractive tradeoff between accuracy and hardware efficiency for specific neural network (NN) workloads. This work presents the first multi-mode gain-cell (GC) CIM macro capable of processing MX, integer (INT), and FP multiply-and-accumulate (MAC) operations with high energy efficiency (EEF) and area efficiency (AEF). The proposed macro employs four important innovations: 1) a multi-mode input processing unit (M2-IPU) with SS-variance-aware MAC flow (SS-VAF) for SS processing and SS alignment within the CIM macro to reduce system-to-CIM data transfer and compute energy; 2) a pattern-aware hybrid adder tree (PAH-ADT), which improves EEF and AEF by optimizing the common input patterns; 3) an accumulation-aware data flow (A2-DF) that adjusts the write path based on accumulation size to reduce data transfer energy; and 4) a 3.xT GC, which boosts data retention time (DRT) by increasing parasitic capacitance without additional area overhead. A 16-nm FinFET 216-kb MX–INT–FP multi-mode GC-CIM macro achieved 133.5 TFLOPS/W for MX-MAC with MXINT8 input, MXINT8 weight, and FP32 output; and 91.9 TFLOPS/W for FP-MAC with BF16 input, BF16 weight, and FP32 output.
- 22 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 171-183
A 22 nm Floating-Point ReRAM Compute-in-Memory Macro Using Residue-Shared ADC for AI Edge Device
Artificial intelligence (AI) edge devices increasingly require the enhanced accuracy of floating-point (FP) multiply-and-accumulate (MAC) operations as well as nonvolatile on-chip memory to minimize the movement of weight data in power-off mode. Designing non-volatile compute-in-memory (nvCIM) macros for FP operations imposes several challenges, including: 1) a tradeoff between inference accuracy and weight bit-width following pre-alignment; 2) long computing latency and high energy consumption; 3) large cell array current during computation; and 4) high multi-bit readout energy consumption. In this study, we devised four schemes to address these issues, including: 1) a kernel-wise weight pre-alignment (K-WPA); 2) a rescheduled multi-bit input compression (RS-MIC); 3) HRS-favored dual-sign-bit (HF-DSB); and 4) residue-shared analog-to-digital converter (RS-ADC). A 16 Mb resistive random access memory (ReRAM) nvCIM macro fabricated for FP operations using foundry-provided ReRAM (22 nm CMOS technology) achieved an efficiency of 34.2 TFLOPS/W under BF16-input, BF16-weight, and FP32-output and 31.4 TFLOPS/W under FP16-input, FP16-weight, and FP32-output.
- 22 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 171-183
A 22 nm Floating-Point ReRAM Compute-in-Memory Macro Using Residue-Shared ADC for AI Edge Device
Artificial intelligence (AI) edge devices increasingly require the enhanced accuracy of floating-point (FP) multiply-and-accumulate (MAC) operations as well as nonvolatile on-chip memory to minimize the movement of weight data in power-off mode. Designing non-volatile compute-in-memory (nvCIM) macros for FP operations imposes several challenges, including: 1) a tradeoff between inference accuracy and weight bit-width following pre-alignment; 2) long computing latency and high energy consumption; 3) large cell array current during computation; and 4) high multi-bit readout energy consumption. In this study, we devised four schemes to address these issues, including: 1) a kernel-wise weight pre-alignment (K-WPA); 2) a rescheduled multi-bit input compression (RS-MIC); 3) HRS-favored dual-sign-bit (HF-DSB); and 4) residue-shared analog-to-digital converter (RS-ADC). A 16 Mb resistive random access memory (ReRAM) nvCIM macro fabricated for FP operations using foundry-provided ReRAM (22 nm CMOS technology) achieved an efficiency of 34.2 TFLOPS/W under BF16-input, BF16-weight, and FP32-output and 31.4 TFLOPS/W under FP16-input, FP16-weight, and FP32-output.
- 15 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 158-170
An Integer-Floating-Point Dual-Mode Gain-Cell Computing-in-Memory Macro for Advanced AI Edge Chips
This article presents a novel integer-floating-point (INT-FP) gain-cell (GC)-computing-in-memory (CIM) structure for high-precision multiply-and-accumulate (MAC) operations with high computational flexibility, energy efficiency, and inference accuracy. The proposed device employs: 1) a dual-mode zone-based input processing scheme (ZB-IPS) aimed at eliminating exponent subtraction in order to enhance energy and area efficiency (AEF); 2) a dual-mode local computing cell (DM-LCC) to reuse exponent addition as an adder tree stage for INT-MAC to enhance AEF in both INT and floating-point (FP) modes; and 3) a stationary-based two-port GC array (SB-TP-GCA) to enable concurrent data updates and computation while reducing system-to-CIM and internal data accesses to improve energy efficiency. A 16-nm FinFET 108-kb GC-CIM macro fabricated using 4T gain cells (GCs) achieved energy efficiency of 99.5 TOPS/W in INT-MAC operations involving 128 accumulations of 8b-input, 8b-weight, and 23b-output; and 46.4 TFLOPS/W in FP-MAC operations involving 64 accumulations of BF16-input, BF16-weight, and FP32-output.
- 15 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Solid-State Circuits
- Vol. 60  (1) , 158-170
An Integer-Floating-Point Dual-Mode Gain-Cell Computing-in-Memory Macro for Advanced AI Edge Chips
This article presents a novel integer-floating-point (INT-FP) gain-cell (GC)-computing-in-memory (CIM) structure for high-precision multiply-and-accumulate (MAC) operations with high computational flexibility, energy efficiency, and inference accuracy. The proposed device employs: 1) a dual-mode zone-based input processing scheme (ZB-IPS) aimed at eliminating exponent subtraction in order to enhance energy and area efficiency (AEF); 2) a dual-mode local computing cell (DM-LCC) to reuse exponent addition as an adder tree stage for INT-MAC to enhance AEF in both INT and floating-point (FP) modes; and 3) a stationary-based two-port GC array (SB-TP-GCA) to enable concurrent data updates and computation while reducing system-to-CIM and internal data accesses to improve energy efficiency. A 16-nm FinFET 108-kb GC-CIM macro fabricated using 4T gain cells (GCs) achieved energy efficiency of 99.5 TOPS/W in INT-MAC operations involving 128 accumulations of 8b-input, 8b-weight, and 23b-output; and 46.4 TFLOPS/W in FP-MAC operations involving 64 accumulations of BF16-input, BF16-weight, and FP32-output.