IEEE Journal of Solid-State Circuits

A 0.29 pJ/b 5.27 Tb/s/mm UCIe Advanced Package Link With 2.5-D CoWoS Interposer and a 0.52 pJ/b 0.448 Tb/s/mm UCIe Standard Package Link With Organic Substrate in 3 nm FinFET

A 0.29 pJ/b 5.27 Tb/s/mm UCIe Advanced Package Link With 2.5-D CoWoS Interposer and a 0.52 pJ/b 0.448 Tb/s/mm UCIe Standard Package Link With Organic Substrate in 3 nm FinFET 150 150

Abstract:

This work presents two die-to-die (D2D) wireline transceivers, one compliant with the UCIe advanced package (UCIe-AP) and the other with the UCIe standard package (UCIe-SP) standard, developed in 3 nm FinFET. The Universal Chiplet interconnect express (UCIe)-AP link has 64 RX and 64 TX data lanes in one PHY module and …

View on IEEE Xplore

HUTAO: A Reconfigurable Homomorphic Processing UniT With Cache-Aware Operation Scheduling

HUTAO: A Reconfigurable Homomorphic Processing UniT With Cache-Aware Operation Scheduling 150 150

Abstract:

Fully homomorphic encryption (FHE) enables privacy-preserving machine learning (PPML) at the cost of intensive computational overhead, which necessitates the use of domain-specific accelerators. To achieve comprehensive support for leveled FHE, this article presents a reconfigurable multi-scheme FHE processor that supports both client-side encryption/decryption and server-side evaluation. First, a reconfigurable …

View on IEEE Xplore

A 256-Point FFT Using Analog Floating-Point Computation With Post-Silicon Tuning

A 256-Point FFT Using Analog Floating-Point Computation With Post-Silicon Tuning 150 150

Abstract:

This article presents a 256-point fast Fourier transform (FFT) processor based on an analog floating-point computation framework. A novel floating-point analog multiply-and-accumulate (FP-AMAC) unit encodes mantissas as pulse widths with a 4-bit digital exponent and operates through integrate, pulse, and reset stages with accuracy enhancements, including node pre-charge and comparator …

View on IEEE Xplore

A 40-nm Programmable Heterogeneous SoC With 5.625/0.85 MB RRAM/SRAM for Accelerating Neuro-Symbolic AI Models

A 40-nm Programmable Heterogeneous SoC With 5.625/0.85 MB RRAM/SRAM for Accelerating Neuro-Symbolic AI Models 150 150

Abstract:

Neuro-symbolic (NeSy) artificial intelligence (AI) integrates neural learning with symbolic reasoning to enable data-efficient, interpretable, and generalizable intelligence, making it a promising paradigm for human-like cognition. However, the heterogeneous and dynamic execution patterns of NeSy models pose fundamental challenges to conventional AI hardware, which is typically optimized for dense matrix …

View on IEEE Xplore

A 77-fJ/bit 8-Gbps Adaptive-Voltage-Compatible Self-Timed Die-to-Die Link for 2.5-D and 3-D Interconnect in 3 nm

A 77-fJ/bit 8-Gbps Adaptive-Voltage-Compatible Self-Timed Die-to-Die Link for 2.5-D and 3-D Interconnect in 3 nm 150 150

Abstract:

This work presents a self-timed die-to-die link that serializes four data bits per pin for 2.5-D, or 3-D interconnects using a standard adaptive digital clock and voltage supply. The link achieves 8 Gbps of per-pin bandwidth with a latency of one cycle, energy efficiency of 77 fJ/b, and bandwidth density of 44…

View on IEEE Xplore

A Low-Reference-Spur Injection-Locked Clock Multiplier Using Sub-Sampling Frequency Tracking Loop and Injection Pulse Timing Calibrator

A Low-Reference-Spur Injection-Locked Clock Multiplier Using Sub-Sampling Frequency Tracking Loop and Injection Pulse Timing Calibrator 150 150

Abstract:

This article presents an injection-locked clock multiplier (ILCM) achieving the low-reference spur (spur ${}_{\mathrm {REF}}$ ) with minimal overhead of a calibrator. To remove the dominant sources of frequency error, which are frequency drift ( $f_{\mathrm {DF}}$ ), phase offset ( $\varPhi _{\mathrm {OS}}$ ), and injection-induced phase error ( $\varPhi _{\mathrm {INJ}}$ ), the ILCM …

View on IEEE Xplore

A Modular Ring-Oscillator Array Chip for Accurate Stress Testing of CMOS Aging Mechanisms

A Modular Ring-Oscillator Array Chip for Accurate Stress Testing of CMOS Aging Mechanisms 150 150

Abstract:

Ring-oscillator (RO) circuits have historically been used to characterize the performance of CMOS technologies, as they can easily expose both process variability and aging through a straightforward circuit structure. ROs are widely employed to study degradation mechanisms such as bias temperature instability (BTI) and hot carrier degradation (HCD), which progressively …

View on IEEE Xplore

A 4.21-to-15.18 GHz Magnetically Reconfigurable and Fully Symmetrical Quad-Core Quad-Mode VCO

A 4.21-to-15.18 GHz Magnetically Reconfigurable and Fully Symmetrical Quad-Core Quad-Mode VCO 150 150

Abstract:

This work proposes a wideband low phase noise (PN) quad-core quad-mode voltage-controlled oscillator (VCO) which features magnetically reconfigurable (MR) technique and fully symmetrical architecture. The MR technique, realized through a transformer-based tank integrating an embedded switched inductor, leverages varying core excitation currents to achieve four distinct equivalent inductances, thereby enabling …

View on IEEE Xplore

Cryogenic CMOS Microwave Signal Selector Using Dual-Stage Injection-Locked Oscillator for Frequency-Multiplexed Qubit Control System

Cryogenic CMOS Microwave Signal Selector Using Dual-Stage Injection-Locked Oscillator for Frequency-Multiplexed Qubit Control System 150 150

Abstract:

Cryogenic CMOS (cryo-CMOS) quantum bit (qubit) control circuits are expected to overcome the interconnect complexity problem in large-scale quantum computers. However, since each qubit has a unique frequency for control, many power-hungry oscillators are needed to generate the frequencies at the cryogenic temperature stage of a refrigerator. In this article, …

View on IEEE Xplore