检索结果-内蒙古大学图书馆

ETBench: Characterizing Hybrid Vision Transformer Workloads Across Edge Devices

IEEE Transactions on Computers 2025年第6期74卷 1857-1871页

作者： Zhou, Yingkun Tian, Zhengshuyuan Yang, Wenhao Zhang, Tingting Ye, Jinpeng Han, Chenji Liu, Tianyi Zhang, Fuxin Institute of Computing Technology CAS State Key Lab of Processors Beijing100190 China University of Chinese Academy of Sciences Beijing100049 China Loongson Technology Company Ltd. Beijing100095 China Institute of Computing Technology CAS Beijing100190 China University of Texas at San Antonio San AntonioTX78249 United States

Lightweight Convolution and Vision Transformer hybrid models have increasingly dominated the frontiers of deep learning (DL) on edge devices;however, to the best of our knowledge, no prior work has provided comprehensive evaluation on hybrid models’ performance and analyzed their characteristics by diving deep into the edge ecosystem with diversified modern DL inference engines and heterogeneous hardware. This paper proposes a comprehensive open-source benchmark suite, ETBench, to allow power-efficiency, performance and accuracy assessment for state-of-the-art (SOTA) hybrid models across 11 most widely-used DL engines deployed on diverse edge devices. After building ETBench that satisfies 6 design requirements proposed in our work, we conduct extensive experiments on 14 devices including 19 CPUs, 11 GPUs and 5 NPUs, and obtain benchmark results from all deployment scenarios (combinations of models, quantization formats, software engines, and hardware platforms). Valuable observations and insightful implications are finally summarized. For example, within current DL engines, the INT8 quantization is significantly underperformed in terms of accuracy and speed against FP16 for hybrid models. Overall, ETBench serves as a collaborative platform that assists model architects in better evaluating their models and makes it possible for future co-optimizations of DL engines and hardware accelerators. © 1968-2012 IEEE.

关键词： Open source software

来源：评论

学校读者我要写书评

暂无评论

Metal-organic polyhedra as dynamic cross-linker for recyclable mixed matrix membranes with promising dimension stability

引用

science china Chemistry 2025年第2期68卷 705-713页

作者： Linjie Lan Mingxin Zhang Haiyan Xiao Wei Liu-Fu Yuyan Lai Jiadong Chen Jia-Fu Yin Panchao Yin State Key Laboratory of Luminescent Materials and Devices&South China Advanced Institute for Soft Matter Science and Technology Guangdong Basic Research Center of Excellence for Energy and Information Polymer MaterialsSouth China University of TechnologyGuangzhou 510641China State Key Laboratory of Marine Resource Utilization in South China Sea Hainan Provincial Key Lab of Fine ChemistrySchool of Chemical Engineering and TechnologyHainan UniversityHaikou 570228China

Mixed matrix membranes(MMMs)are designed by incorporating microporous inorganic fillers into polymer matrix for the integration of inorganics’excellent gas separation performance with polymers’promising ***,due to the complex interphase interaction,the prevention of long-term aging is still challenging and the poor processability/recyclability of MMMs hinders their extensive ***,a coordination assembly protocol is developed for the fabrication of MMMs with both excellent processability/recyclability upon thermal annealing and promising dimension stability even under high gas *** bearing isophthalic acid(IPA)as side chains are applied to coordinate with Cu^(2+)and the assembled 2.5 nm microporous metal-organic polyhedra(MOP),{Cu_(24)IPA_(24)},serves as multi-functional cross-linkers for the polymer network *** MOPs’well-defined microporosity grants MMMs the capability for O_(2)/N_(2) separation while their activated dynamics at high temperatures(>120℃)endows polymer networks with excellent *** contrast to conventional supramolecular units,the MOPs possess unique logarithmic ligand-exchange kinetics and they show nearly frozen bonding dynamics at room temperature,ensuring promising dimensional stability to overcome the long-term aging and high gas pressure *** work not only provides fundamental understanding of the unique relaxation dynamics of complex systems with intertwined dynamics of supramolecular bonds and polymer chains,but also paves new avenues for the fabrication of robust and recyclable MMMs and elastomers.

关键词： dynamic bonding metal-organic polyhedra mixed matrix membrane polymer recycling gas separation

来源：评论

学校读者我要写书评

暂无评论

Integrated multi-scale synchrotron radiation-technology studies on AlPO_(4)-coating modification mechanism in lithium-rich manganese-based cathode

引用

Nuclear science and Techniques 2025年第5期36卷 30-42页

作者： Zhong-Qin Dai Huan Chen Zhao-Yin Wen School of Physical Science and Technology ShanghaiTech UniversityShanghai 201210China The State Key Lab High-Performance Ceram&Superfine Shanghai Institute of CeramicsChinese Academy of SciencesShanghai 200050China University of Chinese Academy of Science Beijing 100049China

Lithium-and manganese-rich(LMR)oxide cathode materials are among the most attractive candidates for next-generation energy-storage materials owing to their anomalous ***,severe Mn dissolution that occurs during long-term cycling,which leads to capacity loss,hinders their application *** this study,nanoscale AlPO_(4)-coated Li_(1.2)Ni_(0.13)Co_(0.13)Mn_(0.54)O_(2)(LMR@APO)with significantly enhanced electrochemical performance is successfully synthesized using a simple and effective sol–gel method to mitigate Mn dissolution and suppress local structural distortion at high *** of the complex evolution of the structure and oxidation state of LMR materials during electrochemical cycling,observing and analyzing them using traditional single characterization methods may be ***,we combine various synchrotron-based characterization techniques to conduct a detailed analysis of the electronic and coordination structures of the cathode material from the surface to the ***-based hard and soft X-ray spectroscopies are integrated to investigate the differences in O and Mn evolution between the surfaces and bulk of the *** synchrotron-based transmission X-ray microscopy combined with X-ray near-edge absorption-structure technology is utilized to visualize the two-dimensional nanometer-scale reactivity of the LMR *** AlPO_(4)-coating layer can stabilize the surface structure of the LMR material,effectively alleviating irreversible oxygen release on the surface and preventing the dissolution of Mn^(2+)at the interface caused by side reactions after a long ***,the spatial reaction uniformity of Mn is enhanced by the AlPO_(4)-coating layer,and rapid capacity decay caused by Mn deactivation is *** AlPO_(4)-coating method is a facile modification strategy for high-performance LMR materials.

关键词： Synchrotron radiation X-ray absorption fine structure X-ray imaging Amorphous coating layer Lithium-and manganese-rich(LMR)cathode

来源：评论

学校读者我要写书评

暂无评论

Explainable and Layout-Aware Timing Prediction 24

Explainable and Layout-Aware Timing Prediction

引用

43rd International Conference on Computer-Aided Design, ICCAD 2024

作者： Lyu, Zhengyang Li, Xiaqing Du, Zidong Guo, Qi University of Science and Technology of China Hefei China State Key Lab of Processors Institute of Computing Techology CAS Beijing China Shanghai Innovation Center for Processor Technologies Shanghai China Ministry of Education Key Lab for Big Data and AI in Transportation Beijing Jiaotong University Beijing China

ISBN: (纸本)9798400710773

Accurate and fast timing prediction at early design stages is crucial for achieving timing closure in very-large-scale integration (VLSI) design. Machine learning (ML) based approaches have been widely adopted for timing prediction to pursue both of accuracy and speed. Unfortunately, these approaches fail to fully exploit the layout features which can significantly impact the timing metrics, thus leading to great accuracy degradation. In this paper, we propose LE-Timing, an end-to-end explainable and layout-aware timing prediction approach that can achieve better accuracy by leveraging important layout information. The key insight is that crucial information about coupling capacitance between routing wires can significantly impact timing metrics, and fully exploiting this layout information in the timing prediction model can substantially improve its accuracy. However, efficiently leveraging layout features while maintaining accuracy and interpretability poses three key challenges: node-level layout-graph integration, interpretability and training efficiency. To address these, we further propose three key technologies specially designed for LE-Timing correspondingly, including a layout-encoding net model, an explainable cell model, and an explicit-message-passing scheme. Experimental results on 14 open-source designs with advanced 7-nm technology node demonstrate that LE-Timing achieves R2 score 0.92 for arrival time prediction and represents strong interpretability. © 2024 Copyright is held by the owner/author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

NeuVSA: A Unified and Efficient Accelerator for Neural Vector Search 31

NeuVSA: A Unified and Efficient Accelerator for Neural Vecto...

引用

31st IEEE International Symposium on High Performance Computer Architecture, HPCA 2025

作者： Yuan, Ziming Dai, Lei Li, Wen Zhang, Jie Liang, Shengwen Wang, Ying Liu, Cheng Li, Huawei Li, Xiaowei Guo, Jiafeng Wang, Peng Chen, Renhai Zhang, Gong Shanxi University School of Computer and Information Technology China Peking University School of Computer Science China Institute of Computing Technology CAS State Key Lab of Processors China University of Chinese Academy of Sciences China Zhongguancun National Laboratory Beijing China Institute of Computing Technology CAS Key Lab of Network Data Science and Technology China Huawei Technologies Co. Ltd. China

ISBN: (纸本)9798331506476

Neural Vector Search (NVS) has exhibited superior search quality over traditional key-based strategies for information retrieval tasks. An effective NVS architecture requires high recall, low latency, and high throughput to enhance user experience and cost-efficiency. However, implementing NVS on existing neural network accelerators and vector search accelerators is sub-optimal due to the separation between the embedding stage and vector search stage at both algorithm and architecture levels. Fortunately, we unveil that Product Quantization (PQ) opens up an opportunity to break separation. However, existing PQ algorithms and accelerators still focus on either the embedding stage or the vector search stage, rather than both simultaneously. Simply combining existing solutions still follows the beaten track of separation and suffers from insufficient parallelization, frequent data access conflicts, and the absence of scheduling, thus failing to reach optimal recall, latency, and throughput. To this end, we propose a unified and efficient NVS accelerator dubbed NeuVSA based on algorithm and architecture co-design philosophy. Specifically, on the algorithm level, we propose a learned PQ-based unified NVS algorithm that consolidates two separate stages into the same computing and memory access paradigm. It integrates an end-to-end joint training strategy to learn the optimal codebook and index for enhanced recall and reduced PQ complexity, thus achieving smoother acceleration. On the architecture level, we customize a homogeneous NVS accelerator based on the unified NVS algorithm. Each sub-accelerator is optimized to exploit all parallelism exposed by unified NVS, incorporating a structured index assignment strategy and an elastic on-chip buffer to alleviate buffer conflicts for reduced latency. All sub-accelerators are coordinated using a hardware-aware scheduling strategy for boosted throughput. Experimental results show that the joint training strategy improves recall

关键词： Network embeddings

来源：评论

学校读者我要写书评

暂无评论

Asymmetric charge distribution boosts hydrogen evolution performance in two-dimensional MoO_(2)/MoS_(2)step heterostructure

引用

Carbon Energy 2025年第2期7卷 172-182页

作者： Mengke Kang Xiang Zhang Jingyi Wang Wen Li Tianyu Xue Kun Zhai Jianyong Xiang Anmin Nie Yingchun Cheng Zhongyuan Liu Center for High Pressure Science State Key Lab of Metastable Materials Science and TechnologyYanshan UniversityQinhuangdaoChina Key Laboratory of Flexible Electronics&Institute of Advanced Materials Jiangsu National Synergetic Innovation Center for Advanced MaterialsNanjing Tech UniversityNanjingChina Department of Chemistry Illinois Institute of TechnologyChicagoIllinoisUSA

Step heterostructures are predicted to hold a profound catalytic performance because of the rearranged electronic structure at their ***,limitations in the morphology of heterostructures prepared by hydrothermal reactions or molten salt-assisted strategies make it challenging to directly assess charge distribution and evaluate a single interface's hydrogen evolution reaction(HER)***,we prepared two-dimensional MoO_(2)/MoS_(2) step heterostructures with a large specific surface area by the chemical vapor deposition *** Kelvin probe force microscopy and electrical transport measurement verified the asymmetric charge distribution at a single *** fabricating a series of micro on-chip electrocatalytic devices,we investigate the HER performance for a single interface and confirm that the interface is essential for superior catalytic *** experimentally confirmed that the enhancement of the HER performance of step heterostructure is attributed to the asymmetric charge distribution at the *** work lays a foundation for designing highly efficient catalytic systems based on step heterostructures.

关键词： charge distribution chemical vapor deposition hydrogen evolution reaction interface step heterostructure

来源：评论

学校读者我要写书评

暂无评论

Latency Insensitivity Testing for Dataflow HLS Designs 25

Latency Insensitivity Testing for Dataflow HLS Designs

引用

33rd ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2025

作者： Cheng, Jianyi Wang, Lianghui Jiang, Zijian Bao, Yungang Shi, Kan School of Informatics The University of Edinburgh Edinburgh United Kingdom Institute of Computing Technology Chinese Academy of Sciences University of Science and Technology of China Beijing China State Key Lab of Processors Institute of Computing Technology Chinese Academy of Sciences University of Chinese Academy of Sciences Beijing China

ISBN: (纸本)9798400713965

Dataflow high-level synthesis (HLS) tools automatically map a high-level software program to a dataflow hardware design. When testing the design, the HLS tool takes a testing function written in the same software language and translates it into the corresponding hardware testbench for cycle-accurate simulation. However, the generated testbench only follows a fixed schedule to feed inputs and collect outputs, without considering external dynamic behaviors of the test data (also known as latency insensitivity testing). Overlooking these test cases could miss bugs introduced by the designer that occur in certain events, such as late input arrival and output backpressure, leading to higher debugging costs. In this paper, we present an HLS testing toolflow named PEDIA that automatically generates testbenches to detect latency insensitivity bugs. We expand the test space from varying input values to varying input delays. PEDIA automatically determines an efficient test space of input delays based on the HLS schedule for better scalability;and tests the HLS design on a parallel hardware testing platform for high performance. Over two representative sets of existing benchmarks, we show that PEDIA can help identify potential bugs missed by existing approaches and achieves an average speedup of 4.53× on running 1500 test cases. © 2025 Copyright held by the owner/author(s).

关键词： Software testing

来源：评论

学校读者我要写书评

暂无评论

ETPG: Efficient Transition Fault Simulation via Dual-Strategy Pattern Parallelism and Gate Restructuring 25

ETPG: Efficient Transition Fault Simulation via Dual-Strateg...

引用

30th Asia and South Pacific Design Automation Conference, ASP-DAC 2025

作者： Wang, Mingjun Wang, Hui Liu, Zizhen Gu, Feng Mu, Jianan Tang, Jiaping Gao, Jun Li, Huawei Ye, Jing Li, Xiaowei State Key Lab of Processors Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Castest Co. Ltd. Beijing China

ISBN: (纸本)9798400706356

With the advancement of integrated circuit (IC) technology, the sensitivity to delay defects has significantly increased, rendering Transition Fault (TF) testing crucial for ensuring chip quality. However, as the complexity of IC designs increases, existing pattern parallelization methods are not flexible in detecting multi-cycle faults. In addition, the growing demand for simulation memory exacerbates inefficient memory access, becoming another critical bottleneck. This paper introduces ETPG (Efficient Transition fault simulation via dual-strategy Pattern parallelism and Gate restructuring), a novel TF simulation algorithm based on multi-dimensional optimization. The key innovations include an adaptive dual-strategy pattern parallel strategy that dynamically optimizes parallelization based on test pattern characteristics, enhancing efficiency and multi-cycle fault detection capability;a dual-dimension gate restructuring method that optimizes memory storage order, significantly reducing memory access time, particularly beneficial for large-scale circuits;and a collaborative mechanism between pattern processing and circuit storage optimization, achieving comprehensive performance improvements at both algorithmic and memory access levels. Experimental results demonstrate ETPG's significant performance improvements across various circuit scales, particularly for larger circuits. Compared to the synopsys commercial tool testmax (TMAX), ETPG achieves average speedups of 2.846× for circuits below 100k gates and 4.428× for circuits above 100k gates. © 2025 Institute of Electrical and Electronics Engineers Inc.. All rights reserved.

关键词： Delay circuits

来源：评论

学校读者我要写书评

暂无评论

Improving prompt tuning-based software vulnerability assessment by fusing source code and vulnerability description

引用

Automated Software Engineering 2025年第2期32卷 1-29页

作者： Wang, Jiyu Chen, Xiang Pei, Wenlong Yang, Shaoyu School of Artificial Intelligence and Computer Science Nantong University Jiangsu Nantong China State Key Lab. for Novel Software Technology Nanjing University Jiangsu Nanjing China

To effectively allocate resources for vulnerability remediation, it is crucial to prioritize vulnerability fixes based on vulnerability severity. With the increasingnumber of vulnerabilities in recent years, there is an urgent need for automated methods for software vulnerability assessment (SVA). Most of the previous SVA studies mainly rely on traditional machine learning methods. Recently, fine-tuning pre-trained language models has emerged as an intuitive method for improving performance. However, there is a gap between pre-training and fine-tuning, and their performance heavily depends on the dataset’s quality of the downstream task. Therefore, we propose a prompt tuning-based method PT-SVA. Different from the fine-tuning paradigm, the prompt-tuning paradigm involves adding prompts to make the training process similar to pre-training, thereby better adapting to downstream tasks. Moreover, previous research aimed to automatically predict severity by only analyzing either the vulnerability descriptions or the source code of the vulnerability. Therefore, we further consider both types of vulnerability information for designing hybrid prompts (i.e., a combination of hard and soft prompts). To evaluate PT-SVA, we construct the SVA dataset based on the CVSS V3 standard, while previous SVA studies only consider the CVSS V2 standard. Experimental results show that PT-SVA outperforms ten state-of-the-art SVA baselines, such as by 13.7% to 42.1% in terms of MCC. Finally, our ablation experiments confirm the effectiveness of PT-SVA’s design, specifically in replacing fine-tuning with prompt tuning, incorporating both types of vulnerability information, and adopting hybrid prompts. Our promising results indicate that prompt tuning-based SVA is a promising direction and needs more follow-up studies. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Bi-modal information Hybrid prompt Pre-trained language model Prompt tuning Software vulnerability assessment

来源：评论

学校读者我要写书评

暂无评论

A 316 stainless steel rod impacts with a rigid flat:Theory,experiment,and numerical simulation

引用

science china(Physics,Mechanics & Astronomy) 2025年第5期68卷 164-182页

作者： Yifan Wang Xuan Ye Hao Yan Tao Wang Guangyan Huang Zhuo Zhuang State Key Laboratory of Explosion Science and Safety Protection Beijing Institute of TechnologyBeijing 100081China Institute of Nuclear and New Energy Technology Tsinghua UniversityBeijing 100084China Applied Mechanics Lab. School of Aerospace EngineeringTsinghua UniversityBeijing 100084China

A slender rod suffers global vibration in *** this study,we present the experimental,numerical,and theoretical studies of the axial responses of a 316 stainless steel rod during vertical impact with a rigid *** the contact models and the one-dimensional(1D)wave equation,we first develop a semi-analytical vertical impact model for the rods based on a unified theoretical framework,which considers different geometries of the impacting end including the hemispherical nose,the truncated conical nose,and the flat ***,we perform free-drop experiments on these rods and numerical simulations to verify the theoretical *** results show that the strain-rate effect hardens the rod nose and should not be ignored even at a velocity as low as a few meters per *** the proposal of a dynamic correction factor to adjust the quasi-static contact model,the theoretical,numerical,and experimental results agree well with one ***,the threedimensional(3D)FEM simulations show that the slight deviations between the experimental and the predicted results are due to the slight obliqueness of the rods in the ***,we leverage the theoretical tool and FEM simulations to compare the mechanical responses of rods with different impacting ends,and suggestions about the selection of rod noses are obtained.

关键词： vertical impact slender rod variable separation contact-impact force drop tests

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：