Yifan He

Yifan He

Auto-generated profile, based on publicly available information

Yifan He

Department of Electronic Engineering, Tsinghua University, Beijing, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 June 2025 in IEEE Solid-State Circuits Magazine
Verified
Journal Article
Closed access
YHYifan He
Yifan He
XHXiaofeng Hu
Xiaofeng Hu
HJHongyang Jia
Hongyang Jia
JSJae-Sun Seo
Jae-Sun Seo
Compute-in-memory (CIM) has shown great potential in efficiently processing high-dimensional data over traditional von Neumann architectures, becoming a candidate computing fabric for next-generation AI. This has motivated the rapid development of CIM prototypes and deployments in different approaches, among which SRAM- and eDRAM-based CIM have drawn significant attention due to their flexibility and feasibility. At the time of a decade after the first CIM implementation, it is necessary to review the technical approaches and revisit the new findings behind complicated prototypes. Macro-level innovations such as precise current-based computation and deeply coupled algorithm-circuit co-optimization open up the headroom for efficiency vs. signal-to-noise ratio (SNR) tradeoffs in analog and digital CIM, respectively. Furthermore, diverse architectural configurations integrating CIM macros into systemon- chips have demonstrated scale-out of computing capacity. However, architectural integration of CIM still faces challenges from digital peripherals, memory reloading, and communication. These necessitate hardware-software co-designed mappings and, more importantly, comprehensive and fair evaluation frameworks.

18 August 2025 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
YHYifan He
Yifan He
SFShupei Fan
Shupei Fan
XLXuan Li
Xuan Li
LLLuchang Lei
Luchang Lei
WJWenbin Jia
Wenbin Jia
Yongpan LiuYongpan Liu
Yongpan Liu
Digital circuit is a promising approach to implement computing-in-memory (CIM) architecture for data-intensive applications, such as neural network inference. Previous digital CIM implementations have demonstrated higher throughput and robustness than the analog counterpart, benefiting from technology scaling and full-precision datapaths. However, the traditional digital CIM designs adopt a bulky adder tree as the computing circuit, limiting further improvements in area and energy efficiency. To address the above limitation, this work proposes a novel design methodology that combines the lookup-table (LUT) and high-density embedded dynamic RAM (eDRAM) cells. By storing precomputed weight summations in LUTs, the proposed approach reduces the first two stages of the adder tree. Additionally, the row-level parallelism of the CIM architecture is utilized to amortize the overheads of LUT weight encoding and eDRAM refreshing. The 28-nm prototype demonstrates a peak area efficiency of 16.2 TOPS/mm2 and an energy efficiency of 49.5 TOPS/W, both for 8-bit operations. The proposed eDRAM macro achieves a 2.4 and 0.48 Mb/mm2 density in memory and CIM mode, due to its balanced computing-storage ratio, enabling reconfigurability between computing cores and memory storage.

22 May 2025 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
LLLuchang Lei
Luchang Lei
YZYongqing Zhu
Yongqing Zhu
XZXinhang Zou
Xinhang Zou
ZZZhou Zhang
Zhou Zhang
ZZZian Zhao
Zian Zhao
HJHongyang Jia
Hongyang Jia
This work demonstrates an edge processor featuring programmable and energy-efficient homomorphic encryption (HE) evaluation using in-situ computing (ISC) and in-memory computing (IMC) techniques. HE has been a promising candidate for privacy-preserving computing; however, it is challenged by extreme computation and storage inflation. This has motivated pioneering works on HE hardware in domain-specific accelerator (DSA) forms, whose nature of von Neumann architecture limits their further gains with high-dimensional intermediate ciphertext. The demonstrated processor architecture enables HE evaluation on the edge, addressing intermediate ciphertext’s heavy buffering and swapping overheads. The in-situ HE processing core (ISHEPC) employs tightly coupled dynamic logic and eDRAM bit cells to enable highly dense local storage and parallel programmable computing. An automorphism network implemented at both circuit and architecture levels connects bit cells across ISHEPCs, harnessing chip-level gains with efficient HE data swapping. A silicon prototype integrating eight ISHEPCs with a RISC-V CPU demonstrated in 28-nm CMOS achieves a peak energy efficiency of 3.04 M-NTT/J, a peak throughput of 13.9 k-NTT/s, both with 4096-D 19 bit at 0.8 V, and a local storage density of 2.33 Mb/mm2. This enables 620-k-Op/J CKKS encryption with the same data and voltage configurations. The system demonstration of support-vector-machine (SVM) inference using the CKKS-residue number system (RNS) HE scheme on the LFW dataset achieves 85.5% accuracy with a peak throughput of 2.2 Op/s and a peak energy efficiency of 500 Op/J at 0.8 V.

22 September 2025 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
SYShengzhe Yan
Shengzhe Yan
ZDZhuoyu Dai
Zhuoyu Dai
ZCZhaori Cong
Zhaori Cong
ZGZeyu Guo
Zeyu Guo
YHYifan He
Yifan He
MLMing Liu
Ming Liu
Computing-in-memory (CIM) chips have demonstrated promising energy efficiency for artificial intelligence (AI) applications such as neural networks (NNs), Transformer, and recommendation system (RecSys). However, several challenges still exist. First, a large gap between the macro and system-level CIM energy efficiency is observed. Second, several memory-dominate operations, such as embedding in RecSys and non-linear functions in NN/Transformer, require heavy off-chip access. To solve these challenges, this work presents an energy-efficient CIM-CAM processor for both NN and RecSys applications with three main innovations. First, we propose a reconfigurable Zig-Zag memory access structure, enabling flexible adjacent connections between SRAMs and CIM arrays to minimize SRAM access power. Second, a backbone-systolic (BKS) CIM array is implemented to avoid a long accumulation path with better data reuse between CIM macros. Third, a distribution-aware block-/self-gating (DABG) content-addressable-memory (CAM) is proposed to reduce off-chip access and on-chip search power for the memory-dominate embedding operations in RecSys. The fabricated 28-nm CIM-CAM chip achieves 37.9–81.5 TOPS/W for NN and 12.3–56.1 nJ/request for RecSys and highlights the improvement of system-to-macro efficiency ratio (SMER) by 9.2% over the trendline of the state-of-the-art system-level CIM processors.

1 October 2026 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
YLYaolei Li
Yaolei Li
WGWeichen Gao
Weichen Gao
WJWenbin Jia
Wenbin Jia
XLXinyuan Lin
Xinyuan Lin
SYSongming Yu
Songming Yu
Yongpan LiuYongpan Liu
Yongpan Liu

11 November 2025 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
CTChen Tang
Chen Tang
ZHZongle Huang
Zongle Huang
WWWenxun Wang
Wenxun Wang
XFXiaoyu Feng
Xiaoyu Feng
YHYifan He
Yifan He
Yongpan LiuYongpan Liu
Yongpan Liu
Diffusion-based models have become increasingly prevalent in content creation applications, significantly shaping the future of entertainment, education, and business marketing. In addition to the primary denoising network, these models often incorporate transformer-based modules for customized generation and post-processing stages to ensure high-quality media outputs. However, the diversity of model architectures, coupled with their substantial computational and memory demands, poses a major challenge for efficient inference on personal devices. Facing these challenges, this work proposes CCE, a content creation engine that provides unified acceleration for diffusion-based applications. CCE decomposes activation into outliers and dense features based on the shared task feature, enabling block-wise outlier processing and efficient differential computation for video generation and upscaling. Furthermore, CCE introduces an LUT-based tensor core featuring a bipolar lookup mechanism, which supports mixed-representation computing of different data formats and precision. In addition, a semantic-driven instruction generation unit is developed to compress activation, further enhancing the system’s efficiency adaptively. With these innovations, CCE achieves up to an 84% reduction in memory requirement and $7{\times }$ acceleration. Fabricated using 28-nm CMOS technology, CCE demonstrates $3.58{\times }$ , 28%, and $2{\times }$ higher average energy efficiency compared to prior diffusion, transformer, and super resolution (SR) accelerators, respectively.