Article
Chemistry, Multidisciplinary
Marco A. Giambra, Vaidotas Miseikus, Sergio Pezzini, Simone Marconi, Alberto Montanaro, Filippo Fabbri, Vito Sorianello, Andrea C. Ferrari, Camilla Coletti, Marco Romagnoli
Summary: The article discusses a process for integrating graphene into photonics circuits, demonstrating double-layer graphene electro-absorption modulators with modulation efficiency. The use of chemical vapor deposition and transfer methods ensures reliability of device coverage and performance.
Review
Optics
Lantian Feng, Ming Zhang, Jianwei Wang, Xiaoqi Zhou, Xiaogang Qiang, Guangcan Guo, Xifeng Ren
Summary: This paper reviews the research results and state-of-the-art technologies on the silicon photonic chip for scalable quantum applications, pointing out that some components have already met the requirements for further expansion. It also highlights the challenges ahead and future research directions in on-chip scalable quantum information applications.
PHOTONICS RESEARCH
(2022)
Review
Chemistry, Multidisciplinary
Zheng Gong
Summary: This review summarizes the development of transfer techniques for heterogeneous integration of semiconductor layers and devices, showcasing the advantages and improved performance brought by these technologies for various materials, devices, and systems applications. It also discusses the future research directions of layer transfer and chip transfer techniques.
Article
Optics
Donggyu B. Sohn, Ogulcan E. Orsel, Gaurav Bahl
Summary: Optical isolators based on magneto-optic principles have limitations in integration with photonic circuits, but a non-magnetic alternative utilizing phonon-mediated photonic Autler-Townes splitting shows promise in achieving ultralow insertion loss and high contrast on-chip.
Article
Chemistry, Physical
Zilan Tang, Shula Chen, Dong Li, Xiaoxia Wang, Anlian Pan
Summary: With the increasing demand for fast processing and large data capacity, traditional silicon-based technology is becoming saturated. New strategies, such as using two-dimensional (2D) materials, are being explored. This review focuses on the integration of 2D materials into silicon-based platforms for photonic applications, summarizing their optical properties and discussing the state-of-the-art devices and future challenges.
JOURNAL OF MATERIOMICS
(2023)
Article
Multidisciplinary Sciences
Yilin Xu, Pascal Maier, Matthias Blaicher, Philipp-Immanuel Dietrich, Pablo Marin-Palomo, Wladislaw Hartmann, Yiyang Bao, Huanfa Peng, Muhammad Rodlin Billah, Stefan Singer, Ute Troppenz, Martin Moehrle, Sebastian Randel, Wolfgang Freude, Christian Koos
Summary: This paper presents a novel hybrid ECL utilizing 3D-printed photonic wire bonds as intra-cavity coupling elements to overcome technological challenges. The research demonstrates that photonic wire bonds enable low-loss coupling, leading to wide single-mode tuning range, high side mode suppression ratios, and low intrinsic linewidths.
SCIENTIFIC REPORTS
(2021)
Article
Optics
Mohammad Kaleem, Sajid Nazir, Shaista Jabeen, Hamid Turab Mirza, Shahrukh Agha, Aadil Raza
Summary: This paper presents experimental results of the application of sputtering enhanced Quantum Well Intermixing (QWI) process to modify the bandgap of InGaAsP/InP quantum well laser microstructure. Point defects are created through plasma induced disordering in the process of sputtering Al2O3. Rapid thermal anneal (RTA) process is performed to increase the diffusion rate and achieve quantum well intermixing. The proposed technique provides reliable and fast fabrication of photonic devices at the wafer level.
OPTICS AND LASER TECHNOLOGY
(2023)
Article
Chemistry, Physical
Rui He, Naixin Liu, Yaqi Gao, Renfeng Chen, Siyao Zhang, Hongfeng Yuan, Yiwei Duo, Jintong Xu, Xiaoli Ji, Jianchang Yan, Junxi Wang, Jianguo Liu, Jinmin Li, Tongbo Wei
Summary: This study presents the growth of high-quality AlGaN material and fabrication of monolithic integrated solar-blind LEDs, waveguides, and photodetectors on an AlGaN multi-quantum wells wafer. The waveguides exhibit strong light constraint, and the photodetectors show sufficient photosensitivity to the optical signal traveling along the waveguides, unaffected by ambient lighting.
Article
Optics
Amlan Kusum Mukherjee, Mingjun Xiang, Sascha Preu
Summary: The research introduces a high resistivity silicon-on-insulator based multimodal waveguide topology for the frequency range of 0.5-1.6 THz. This platform, suitable for Terahertz photonic integrated circuits, is mechanically stable and facilitates easy integration of various components.
Article
Optics
Yuansheng Tao, Fenghe Yang, Zihan Tao, Lin Chang, Haowen Shu, Ming Jin, Yan Zhou, Zhangfeng Ge, Xingjun Wang
Summary: Microwave photonics (MWP) is an emerging field that studies the interaction between microwave and lightwave. The recent advances in integrated photonics have provided new opportunities for MWP. In this study, a fully on-chip MWP instantaneous frequency measurement (IFM) system is demonstrated by exploiting hybrid integration of indium phosphide, silicon photonics, and complementary metal-oxide-semiconductor electronics platforms. The unprecedented integration level enhances the compactness, reliability, and performance of the MWP IFM system, and it has been successfully deployed in realistic tasks.
LASER & PHOTONICS REVIEWS
(2022)
Review
Optics
Alexander Dorodnyy, Jasmin Smajic, Juerg Leuthold
Summary: Mie scattering is used to manipulate electromagnetic fields for various applications such as strong resonant enhancement, perfect absorption of radiation, and polarization/wavelength selectivity. Recent applications include light spectrum control, detection, non-linear effects enhancement, and emission. It is also demonstrated that a periodic arrangement of Mie scatterers can lead to a significant absorption enhancement in weakly absorbing layers.
LASER & PHOTONICS REVIEWS
(2023)
Article
Chemistry, Analytical
Lei Wu, Kedong Shang, Tingting Chen, Chengqiang Feng, Tingting Yang, Zhi-Jun Zhao, Bingjun Yu, Linmao Qian
Summary: In this study, channel fabrication with different scales and hybrid features, including channels with single atom layer depth, was successfully achieved using mechano-chemical scanning probe lithography. The selective etching mechanism was explained based on the proposed dissolution model, and the potential applications of nanofluidic devices in cross-scale fabrication were demonstrated.
SENSORS AND ACTUATORS B-CHEMICAL
(2022)
Review
Chemistry, Multidisciplinary
Nasir Ilyas, Jinyong Wang, Chunmei Li, Dongyang Li, Hao Fu, Deen Gu, Xiangdong Jiang, Fucai Liu, Yadong Jiang, Wei Li
Summary: Neuromorphic photonics system based on the principle of biological brain is seen as a potential solution to the bottleneck in classical computing. Optoelectronic synaptic devices using nanostructured materials show promising advantages in light-matter interaction and fast carrier dynamics.
ADVANCED FUNCTIONAL MATERIALS
(2022)
Article
Nanoscience & Nanotechnology
Yikai Su, Yu He, Xuhan Guo, Weiqiang Xie, Xingchen Ji, Hongwei Wang, Xinlun Cai, Limin Tong, Siyuan Yu
Summary: To meet the increasing demand for bandwidth, photonic integrated circuits (PICs) have been widely used in applications requiring high capacity and high-bandwidth density interconnects. However, scaling PICs to meet future petabit per second capacity requirements poses challenges. This study examines the scalability bottlenecks of PICs including guiding materials, dense integration approaches, wide-band optical sources, and high efficiency tunable and modulation devices. Potential solutions to address these challenges are explored. Future PIC development will require significant innovations and technological breakthroughs, as the Moore's law for integrated photonics may have a shorter lifespan compared to that in the microelectronics industry.
Article
Optics
Gaojian Liu, Kaixuan Ye, Okky Daulay, Qinggui Tan, Hongxi Yu, David Marpaung
Summary: In this study, an integrated microwave photonic circuit capable of reconfigurable notch filtering and phase shifting functions in the radio frequency range is demonstrated. The circuit achieves high extinction notch filtering and continuously tunable phase shifting over specific frequency ranges. Furthermore, a linearization technique is implemented to improve the dynamic range of the circuit.
Article
Computer Science, Hardware & Architecture
Byeongho Kim, Jaehyun Park, Eojin Lee, Minsoo Rhu, Jung Ho Ahn
Summary: Personalized recommendation systems are important in industry and the embedding layers within them are memory-intensive. To address performance bottlenecks, a fine-grained near-data processing architecture has been proposed for DRAM, with in-DRAM reduction units at different levels achieving significant performance improvements. Hot embedding-vector replication is also introduced to alleviate load imbalances across reduction units.
IEEE COMPUTER ARCHITECTURE LETTERS
(2021)
Article
Computer Science, Hardware & Architecture
Sunjung Lee, Jaewan Choi, Wonkyung Jung, Byeongho Kim, Jaehyun Park, Hweesoo Kim, Jung Ho Ahn
Summary: Mobile and edge devices are commonly used for inferring CNNs, but existing accelerators are not optimal for the latest CNN models, especially DW-CONV and SE models. This paper proposes a CNN acceleration architecture called MVP, which efficiently processes both compute- and memory-intensive operations with a small area overhead on top of the baseline systolic-array-based architecture.
ACM TRANSACTIONS ON DESIGN AUTOMATION OF ELECTRONIC SYSTEMS
(2022)
Article
Computer Science, Hardware & Architecture
Sungmin Yun, Byeongho Kim, Jaehyun Park, Hwayong Nam, Jung Ho Ahn, Eojin Lee
Summary: Graph Convolutional Network (GCN) models have high accuracy in interpreting graph data, with one of the key components being the aggregation operation. A proposed new architecture, GraNDe, accelerates memory-intensive aggregation operations and achieves a speedup of up to 4.3x on open-graph benchmark datasets compared to baseline systems.
IEEE COMPUTER ARCHITECTURE LETTERS
(2022)
Article
Computer Science, Hardware & Architecture
Yaebin Moon, Wanju Doh, Kwanhee Kyung, Eojin Lee, Jung Ho Ahn
Summary: Tiered memory using DRAM as fast memory and slower-but-larger byte-addressable memory as slow memory is a promising approach to expand main-memory capacity. Proactive demotion schemes are used to demote cold pages to slow memory, even when there is sufficient free space in fast memory. The proposed ADT scheme performs aggressive demotion and promotion by extending the unit of demotion/promotion, reducing fast-memory usage by 29% with only a 2.3% performance drop and outperforming state-of-the-art schemes for tiered memory management.
IEEE COMPUTER ARCHITECTURE LETTERS
(2023)
Article
Computer Science, Hardware & Architecture
Jaewan Choi, Jaehyun Park, Kwanhee Kyung, Nam Sung Kim, Jung Ho Ahn
Summary: Transformer-based generative models utilize attention to summarize input sequences and generate output sequences. However, conventional computing platforms are inefficient in handling attention. To address this issue, we propose AttAcc, which takes advantage of the reuse of KV matrices during summarization and reduces external bandwidth and energy consumption by processing in-memory.
IEEE COMPUTER ARCHITECTURE LETTERS
(2023)
Article
Computer Science, Hardware & Architecture
Hwayong Nam, Seungmin Baek, Minbok Wi, Michael Jaemin Kim, Jaehyun Park, Chihun Song, Nam Sung Kim, Jung Ho Ahn
Summary: The demand for accurate information about the internal structure and characteristics of DRAM is increasing. This paper presents reliable findings on the internal structure and characteristics of DRAM using activate-induced bitflips (AIBs), retention time test, and row-copy operation.
IEEE COMPUTER ARCHITECTURE LETTERS
(2023)
Article
Computer Science, Hardware & Architecture
Hailong Li, Jaewan Choi, Yongsuk Kwon, Jung Ho Ahn
Summary: Transformer-based models are widely used in NLP tasks, but matrix multiplication can be time-consuming. This paper introduces a hardware-friendly approach called tiling singular value decomposition (TSVD) for matrix multiplication, which leverages GPU resources more efficiently and mitigates the loss of important information. The experimental results show that TSVD-matmul achieves significant speedup compared to the SVD approach.
IEEE COMPUTER ARCHITECTURE LETTERS
(2023)
Proceedings Paper
Computer Science, Hardware & Architecture
Jongmin Kim, Gwangho Lee, Sangpyo Kim, Gina Sohn, Minsoo Rhu, John Kim, Jung Ho Ahn
Summary: In this paper, we propose an accelerator called ARK for FHE, which accelerates the bootstrapping operation through runtime data generation and inter-operation key reuse, enabling practical FHE workloads. This approach reduces the size of the working set, maximizes on-chip memory utilization, and effectively handles the heavy computation and data movement overheads of FHE.
2022 55TH ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE (MICRO)
(2022)
Article
Computer Science, Hardware & Architecture
Sunjung Lee, Seunghwan Hwang, Michael Jaemin Kim, Jaewan Choi, Jung Ho Ahn
Summary: This paper investigates the differences in computational performance and memory bandwidth between the CUDA core and the Tensor core in NVIDIA GPUs. Through comparisons and analysis of different generations of Tensor cores, a new method to reduce shared memory traffic is proposed. The experimental results show that inter-warp multicasting significantly improves the performance of deep neural networks.
IEEE TRANSACTIONS ON COMPUTERS
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Sangpyo Kim, Jongmin Kim, Michael Jaemin Kim, Wonkyung Jung, John Kim, Minsoo Rhu, Jung Ho Ahn
Summary: Homomorphic encryption enables secure cloud computation by performing computations on encrypted data. However, the technique of bootstrapping, which allows unlimited operations or fully homomorphic encryption, requires significant additional computation and memory bandwidth. This paper proposes BTS, a hardware accelerator that supports bootstrapping as a first-class citizen, achieving improved execution time through parallel processing elements and deterministic communication patterns.
PROCEEDINGS OF THE 2022 THE 49TH ANNUAL INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE (ISCA '22)
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Michael Jaemin Kim, Jaehyun Park, Yeonhong Park, Wanju Doh, Namhoon Kim, Tae Jun Ham, Jae W. Lee, Jung Ho Ahn
Summary: The Row Hammer (RH) phenomenon has attracted significant attention from the research community due to its security implications. Existing RH-protection schemes have various shortcomings and limitations. This paper introduces Mithril, the first RFM interface-compatible, DRAM-MC cooperative RH-protection scheme, and its optional extension Mithril+. The proposed schemes aim to address the challenges and provide deterministic protection guarantees.
2022 IEEE INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE COMPUTER ARCHITECTURE (HPCA 2022)
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Deok-Jae Oh, Yaebin Moon, Eojin Lee, Tae Jun Ham, Yongjun Park, Jae W. Lee, Jung Ho Ahn
Summary: MaPHeA is a lightweight memory hierarchy-aware profile-guided heap allocation framework applicable to both HPC and embedded systems. It improves application performance by optimizing the allocation of dynamically allocated heap objects with low profiling overhead and without additional user intervention. By identifying frequently accessed heap objects and allocating them to fast DRAM regions, MaPHeA can significantly improve the performance of memory-intensive workloads.
LCTES '21: PROCEEDINGS OF THE 22ND ACM SIGPLAN/SIGBED INTERNATIONAL CONFERENCE ON LANGUAGES, COMPILERS, AND TOOLS FOR EMBEDDED SYSTEMS
(2021)
Proceedings Paper
Computer Science, Hardware & Architecture
Wonkyung Jung, Eojin Lee, Sangpyo Kim, Namhoon Kim, Keewoo Lee, Chohong Min, Jung Hee Cheon, Jung Ho Ahn
2021 IEEE INTERNATIONAL SYMPOSIUM ON PERFORMANCE ANALYSIS OF SYSTEMS AND SOFTWARE (ISPASS 2021)
(2021)
Article
Computer Science, Information Systems
Wonkyung Jung, Eojin Lee, Sangpyo Kim, Jongmin Kim, Namhoon Kim, Keewoo Lee, Chohong Min, Jung Hee Cheon, Jung Ho Ahn
Summary: Homomorphic Encryption (HE) is a popular privacy-preserving approach for cloud computing, with schemes like HE for Arithmetic of Approximate Numbers (HEAAN) gaining popularity due to their support for approximate computations and unlimited arithmetic operations. However, the high computation complexity of HE, especially in ciphertext arithmetic like HE multiplication (HE Mul), has led to a lack of rigorous analysis in accelerating HE and optimizing performance for different parallel processing platforms.
Article
Computer Science, Hardware & Architecture
Hweesoo Kim, Sunjung Lee, Jaewan Choi, Jung Ho Ahn
Summary: The SysAr+ structure proposed in this letter enhances data reuse in the CONV layer without the need for im2col pre-processing, resulting in significant energy consumption reduction and improved performance in ResNet and DenseNet models.
IEEE COMPUTER ARCHITECTURE LETTERS
(2021)