检索结果-内蒙古大学图书馆

IEEE COMPUTER architecture LETTERS 2022年第2期21卷 49-52页

作者： Ma, Rui Georganas, Evangelos Heinecke, Alexander Gribok, Sergey Boutros, Andrew Nurvitadhi, Eriko Univ Texas Austin Elect & Comp Engn Austin TX 78712 USA Intel Corp Santa Clara CA 95054 USA Intel Corp Accelerator Architecture Lab Hillsboro OR 97124 USA

Training state-of-the-art artificial intelligence (AI) models requires scaling to many compute nodes and relies heavily on collective communication operations, such as all-reduce, to exchange the weight gradients between nodes. The overhead of these operations can bottleneck training performance as the number of nodes increases. In this paper, we first characterize the all-reduce operation overhead. Then, we propose a new smart network interface card (NIC) for distributed AI training using field-programmable gate arrays (FPGAs) to accelerate all-reduce operations and optimize bandwidth utilization via data compression. The AI smart NIC frees up the system's compute resources to perform the more compute-intensive tensor operations and increases the overall node-to-node communication efficiency. We build a prototype 6-node AI training system and show that our proposed FPGA-based AI smart NIC enhances overall training performance by 1.6x, with an estimated 2.5x performance improvement at 32 nodes.

关键词： Training Artificial intelligence Field programmable gate arrays Tensors Computational modeling Bandwidth Scalability AI training all-reduce smart NIC FPGA

来源：评论

学校读者我要写书评

暂无评论

Characterization and utilization potential of typical molybdenum tailings in Shaanxi Province, China

引用

ENVIRONMENTAL GEOCHEMISTRY AND HEALTH 2024年第8期46卷 265-265页

作者： Hu, Shentao Xiong, Xiaohu Li, Xuan Wang, Meixia Xu, Dongdong Pan, Aifang Zhou, Weijian Chinese Acad Sci Inst Earth Environm State Key Lab Loess & Quaternary Geol Xian 710061 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China Xian AMS Ctr Shaanxi Prov Key Lab Accelerator Mass Spectrometry Xian 710061 Peoples R China Xian Inst Innovat Earth Environm Res Xian 710061 Peoples R China Xian Univ Architecture & Technol Inst Interdisciplinary & Innovat Res Xian 710055 Peoples R China

Shaanxi Province is located in the most important molybdenum ore district in the world, but a lot of molybdenum tailings have been released, polluting the environment and wasting resources seriously. Taking eleven tailing samples collected at the main molybdenum tailings ponds in Shaanxi Province as the research object, the physical, chemical, and mineralogical characteristics were studied through scanning electron microscope, X-ray fluorescence, X-ray diffraction, inductively coupled plasma mass spectrometer, and others. The ecological risk and utilization potential of molybdenum tailings were investigated through leaching test, geo-accumulation index, potential ecological risk assessment, and other methods. The results demonstrated that the main chemical and mineralogical composition of various molybdenum tailings in Shaanxi Province is similar, and the predominant mineral composition is muscovite, quartz, microcline, and calcite. The potential ecological risk of heavy metals in six molybdenum tailings is high, while Pb and Cd are the main pollution risk elements. Molybdenum tailings contain considerable amounts of critical minerals with huge potential economic value, and molybdenum tailings with high environmental hazards could be converted into a possible source for critical minerals by recovering the critical minerals and repurposing the secondary tailings as an additive or cement substitute. This study provides an innovative idea for the pollution treatment of molybdenum tailings and indicates the prospect of molybdenum tailings as a secondary source for critical minerals.

关键词： Molybdenum tailings Basic characteristics Environmental hazard Utilization potential

来源：评论

学校读者我要写书评

暂无评论

Iron Reduction in Profundal Sediments of Ultraoligotrophic Lake Tahoe under Oxygen-Limited Conditions

引用

ENVIRONMENTAL SCIENCE & TECHNOLOGY 2023年第3期57卷 1529-1537页

作者： Aeppli, Meret Schladow, Geoffrey Pacheco, Juan S. Lezama Fendorf, Scott Stanford Univ Dept Earth Syst Sci Stanford CA 94305 USA Ecole Polytech Fed Lausanne Sch Architecture Civil & Environm Engn CH-1015 Lausanne Switzerland Univ Calif Davis Dept Civil & Environm Engn Davis CA 95616 USA UC Davis Tahoe Environm Res Ctr Incline Village NV 89451 USA SLAC Natl Accelerator Lab Stanford Synchrotron Radiat Lightsource Menlo Pk CA 94025 USA

Increased periods of bottom water anoxia in deep temperate lakes due to decreasing frequency and depth of water column mixing in a warming climate may result in the reductive dissolution of iron minerals and increased flux of nutrients from the sediment into the water column. Here, we assessed the sediment properties and reactivities under depleted oxygen concentrations of Lake Tahoe, a deep ultraoligotrophic lake in the Sierra Nevada mountain range. Using whole-core incubation experiments, we found that a decrease in dissolved oxygen concentration in the top 2 cm of the sediment resulted in an extension of the microbial iron reduction zone from below 4.5 to below 1.5 cm depth. Concentrations of reactive iron generally decreased with sediment depth, and microbial iron reduction seemingly ceased as concentrations of Fe(II) approximated concentrations of reactive iron. These findings suggest that microorganisms preferentially utilized reactive iron and/or iron minerals became less reactive due to mineral transformation and surface passivation. The estimated release of iron mineral-associated phosphorus is not expected to change Lake Tahoe's trophic state but will likely contribute to increased phytoplankton productivity if mixed into surface waters.

关键词： lake sediment iron reduction nutrient release anoxia sediment mineralogy whole-core incubations

来源：评论

学校读者我要写书评

暂无评论

ECON-T and ECON-D: Endcap Concentrator ASICs for the CMS HGCAL

引用

JOURNAL OF INSTRUMENTATION 2025年第3期20卷 C03054-C03054页

作者： Bergamin, G. Ceresa, D. Coko, D. Cummings, G. Gingu, V. Hammer, M. Hirschauer, J. Hoff, J. Kharwadkar, N. Kulis, S. Lupi, M. Mantilla-Suarez, C. Noonan, D. Rubinov, P. Scarfi, S. Shenai, A. Syal, C. Wang, X. Wickwire, R. Wilson, J. Babbar, J. Bam, B. Campbell, A. Klabbers, P. Miao, Y. Francisco, R. Gong, D. Guo, D. Leitao, P. Leroux, P. Moreira, P. Prinzie, J. Yang, D. Ye, J. Fermilab Natl Accelerator Lab Batavia IL 60510 USA European Org Nucl Res CERN CH-1211 Geneva 23 Switzerland Univ Split Fac Elect Engn Mech Engn & Naval Architecture R Boskovica 32 Split Croatia Argonne Natl Lab Lemont IL 60439 USA Baylor Univ Dept Phys Waco TX 76706 USA Univ Alabama Dept Phys Tuscaloosa AL 35401 USA Panjab Univ Dept Phys Sect 14 Chandigarh India Northwestern Univ Dept Phys Evanston IL 60208 USA Katholieke Univ Leuven B-3000 Leuven Belgium Southern Methodist Univ University Pk TX USA

With over 6 million channels, the High Granularity Calorimeter (HGCAL) for the CMS HL-LHC Upgrade presents a unique data challenge. The ECON ASICs provide critical on-detector data reduction for the 40 MHz trigger path (ECON-T) and 750 kHz data acquisition path (ECON-D) of the HGCAL. The ASICs, fabricated in 65 nm CMOS, are rad-tolerant (600 Mrad) with low power consumption (<2.5 mW/channel). This presentation is the first comprehensive description of the ECON designs, first functionality and radiation tests for the ECON-T ASIC, and first results from the full production of 75k ECON-D and ECON-T ASICs.

关键词： Digital electronic circuits Front-end electronics for detector readout Radiation-hard electronics VLSI circuits

来源：评论

学校读者我要写书评

暂无评论

First test results of the HGCAL concentrator ASICs: ECON-T and ECON-D

引用

JOURNAL OF INSTRUMENTATION 2024年第3期19卷 C03050页

作者： Braga, Davide Cummings, Grace Gingu, Cristinel Hirschauer, James Hoff, James Klabbers, Pamela Kharwadkar, Neha Mantilla-Suarez, Cristina Noonan, Daniel Rubinov, Paul Shenai, Alpana Syal, Chinar Wang, Xiaoran Wickwire, Ralph Bergamin, Gianmario Ceresa, Davide Kulis, Szymon Lupi, Matteo Scarfi, Simone Coko, Duje Hammer, Mike Wilson, Jon Bam, Bhim Babbar, Jyoti Campbell, Alex Miao, Yulun Feas, Bram Gong, Datao Guo, Di Yang, Dongxu Prinzie, Jeffrey Ye, Jingbo Leroux, Paul Jozef Moreira, Paulo Leitao, Pedro De Oliveira, Rui Fermilab Natl Accelerator Lab Batavia IL 60510 USA European Org Nucl Res CERN CH-1211 Geneva 23 Switzerland Univ Split Fac Elect Engn Mech Engn & Naval Architecture R Boskovica 32 Split Croatia Argonne Natl Lab Argonne IL 60439 USA Baylor Univ Dept Phys Waco TX 76706 USA Univ Alabama Dept Phys Tuscaloosa AL 35401 USA Panjab Univ Dept Phys Sect 14 Chandigarh India Northwestern Univ Dept Phys & Astron Evanston IL 60208 USA Katholieke Univ Leuven B-3000 Leuven Belgium Southern Methodist Univ University Pk TX USA

With over 6 million channels, the High Granularity Calorimeter for the CMS HL-LHC upgrade presents a unique data transmission challenge. The ECON ASICs provide a critical stage of on -detector data compression and selection for the trigger path (ECON-T) and data acquisition path (ECON-D) of the HGCAL. The ASICs, fabricated in 65 nm CMOS, are radiation tolerant up to 200 Mrad and require low power consumption: < 2.5 mW/sensor-channel per chip. We report on the first functionality and radiation tests for the ECON-D-P1 full -functionality prototype. We present a comparison of single event effect (SEE) cross sections measured for different methods of triple modular redundancy using test results from the ECON-T-P1 full -functionality prototype.

关键词： Front-end electronics for detector readout Radiation-hard electronics

来源：评论

学校读者我要写书评

暂无评论

Large contribution to secondary organic aerosol from isoprene cloud chemistry

引用

SCIENCE ADVANCES 2021年第13期7卷 000-000页

作者： Lamkaddam, Houssni Dommen, Josef Ranjithkumar, Ananth Gordon, Hamish Wehrle, Gunther Krechmer, Jordan Majluf, Francesca Salionov, Daniil Schmale, Julia Bjelic, Sasa Carslaw, Kenneth S. El Haddad, Imad Baltensperger, Urs Paul Scherrer Inst Lab Atmospher Chem CH-5232 Villigen Switzerland Univ Leeds Sch Earth & Environm Leeds LS2 9JT W Yorkshire England Carnegie Mellon Univ Engn Res Accelerator Pittsburgh PA 15213 USA Aerodyne Res Inc Billerica MA 01821 USA Paul Scherrer Inst Bioenergy & Catalysis Lab CH-5232 Villigen Switzerland Ecole Polytech Fed Lausanne Sch Architecture Civil & Environm Engn Lausanne Switzerland

Aerosols still present the largest uncertainty in estimating anthropogenic radiative forcing. Cloud processing is potentially important for secondary organic aerosol (SOA) formation, a major aerosol component: however, laboratory experiments fail to mimic this process under atmospherically relevant conditions. We developed a wetted-wall flow reactor to simulate aqueous-phase processing of isoprene oxidation products (iOP) in cloud droplets. We find that 50 to 70% (in moles) of iOP partition into the aqueous cloud phase, where they rapidly react with OH radicals, producing SOA with a molar yield of 0.45 after cloud droplet evaporation. Integrating our experimental results into a global model, we show that clouds effectively boost the amount of SOA. We conclude that, on a global scale, cloud processing of iOP produces 6.9 Tg of SOA per year or approximately 20% of the total biogenic SOA burden and is the main source of SOA in the mid-troposphere (4 to 6 km).

关键词： Free radicals

来源：评论

学校读者我要写书评

暂无评论

A unique van Hove singularity in kagome superconductor CsV3-xTaxSb5 with enhanced superconductivity

引用

NATURE COMMUNICATIONS 2023年第1期14卷 3819页

作者： Luo, Yang Han, Yulei Liu, Jinjin Chen, Hui Huang, Zihao Huai, Linwei Li, Hongyu Wang, Bingqian Shen, Jianchang Ding, Shuhan Li, Zeyu Peng, Shuting Wei, Zhiyuan Miao, Yu Sun, Xiupeng Ou, Zhipeng Xiang, Ziji Hashimoto, Makoto Lu, Donghui Yao, Yugui Yang, Haitao Chen, Xianhui Gao, Hong-Jun Qiao, Zhenhua Wang, Zhiwei He, Junfeng Univ Sci & Technol China Dept Phys Hefei 230026 Anhui Peoples R China Univ Sci & Technol China CAS Key Lab Strongly coupled Quantum Matter Phys Hefei 230026 Anhui Peoples R China Fuzhou Univ Dept Phys Fuzhou 350108 Fujian Peoples R China Beijing Inst Technol Ctr Quantum Phys Sch Phys Key Lab Adv Optoelect Quantum Architecture & Measu Beijing 100081 Peoples R China Beijing Inst Technol Beijing Key Lab Nanophoton & Ultrafine Optoelect S Beijing 100081 Peoples R China Chinese Acad Sci Beijing Natl Ctr Condensed Matter Phys Beijing 100190 Peoples R China Chinese Acad Sci Inst Phys Beijing 100190 Peoples R China SLAC Natl Accelerator Lab Stanford Synchrotron Radiat Lightsource Menlo Pk CA 94025 USA Univ Sci & Technol China Int Ctr Quantum Design Funct Mat Hefei 230026 Anhui Peoples R China Beijing Inst Technol Mat Sci Ctr Yangtze Delta Reg Acad Jiaxing 314011 Peoples R China

Van Hove singularity (VHS) has been considered as a driving source for unconventional superconductivity. A VHS in two-dimensional (2D) materials consists of a saddle point connecting electron-like and hole-like bands. In a rare case, when a VHS appears at Fermi level, both electron-like and hole-like conduction can coexist, giving rise to an enhanced density of states as well as an attractive component of Coulomb interaction for unconventional electronic pairing. However, this van Hove scenario is often destroyed by an incorrect chemical potential or competing instabilities. Here, by using angle-resolved photoemission measurements, we report the observation of a VHS perfectly aligned with the Fermi level in a kagome superconductor CsV3-xTaxSb5 (x similar to 0.4), in which a record-high superconducting transition temperature is achieved among all the current variants of AV(3)Sb(5) (A = Cs, Rb, K) at ambient pressure. Doping dependent measurements reveal the important role of van Hove scenario in boosting superconductivity, and spectroscopic-imaging scanning tunneling microscopy measurements indicate a distinct superconducting state in this system.

关键词： Electronic properties and materials Superconducting properties and materials

来源：评论

学校读者我要写书评

暂无评论

Inductor Design Optimization Using FEA Supervised Machine Learning 24

Inductor Design Optimization Using FEA Supervised Machine Le...

引用

24th European Conference on Power Electronics and Applications (EPE ECCE Europe)

作者： Cajander, D. Viarouge, I. Viarouge, P. Aguglia, D. HES SO Univ Appl Sci & Arts Western Switzerland Haute Ecole Ingenieurie & Architecture Fribourg ENERGY Inst Blvd Perolles 80 CH-1700 Fribourg Switzerland ELECTROTROTECHOL Selem Inc 2610Rue Gerard Lajoie Quebec City PQ Canada Univ Laval Elect & Comp Eng Dept LEEPCI Lab Quebec City PQ GIK 7P4 Canada CERN European Org Nucl Res Accelerator Syst Dept Elect Power Converter Grp CH-1211 Geneva 23 Switzerland

ISBN: (纸本)9789075815399

An optimal inductor design methodology using dimensioning models derived from Finite Element Analysis (FEA) supervised Artificial Neural Networks (ANN) is presented. The efficiency of such trained ANN dimensioning models in terms of compromise between precision and computing time is demonstrated for the cylindrical inductor topology with air and magnetic material core including saturation.

关键词： Magnetic device Machine learning Deep Learning Design optimization Finite-element analysis Data analysis Neural network Passive component

来源：评论

学校读者我要写书评

暂无评论

WRPN: Wide reduced-precision networks 6

WRPN: Wide reduced-precision networks

引用

6th International Conference on Learning Representations, ICLR 2018

作者： Mishra, Asit Nurvitadhi, Eriko Cook, Jeffrey J. Marr, Debbie Accelerator Architecture Lab Intel Labs

For computer vision applications, prior works have shown the efficacy of reducing numeric precision of model parameters (network weights) in deep neural networks. Activation maps, however, occupy a large memory footprint during both the training and inference step when using mini-batches of inputs. One way to reduce this large memory footprint is to reduce the precision of activations. However, past works have shown that reducing the precision of activations hurts model accuracy. We study schemes to train networks from scratch using reduced-precision activations without hurting accuracy. We reduce the precision of activation maps (along with model parameters) and increase the number of filter maps in a layer, and find that this scheme matches or surpasses the accuracy of the baseline full-precision network. As a result, one can significantly improve the execution efficiency (e.g. reduce dynamic memory footprint, memory bandwidth and computational energy) and speed up the training and inference process with appropriate hardware support. We call our scheme WRPN - wide reduced-precision networks. We report results and show that WRPN scheme is better than previously reported accuracies on ILSVRC-12 dataset while being computationally less expensive compared to previously reported reduced-precision networks. © Learning Representations, ICLR 2018 - Conference Track *** right reserved.

关键词： Chemical activation

来源：评论

学校读者我要写书评

暂无评论

Apprentice: Using knowledge distillation techniques to improve low-precision network accuracy 6

Apprentice: Using knowledge distillation techniques to impro...

引用

6th International Conference on Learning Representations, ICLR 2018

作者： Mishra, Asit Marr, Debbie Accelerator Architecture Lab Intel Labs United States

Deep learning networks have achieved state-of-the-art accuracies on computer vision workloads like image classification and object detection. The performant systems, however, typically involve big models with numerous parameters. Once trained, a challenging aspect for such top performing models is deployment on resource constrained inference systems — the models (often deep networks or wide networks or both) are compute and memory intensive. Low-precision numerics and model compression using knowledge distillation are popular techniques to lower both the compute requirements and memory footprint of these deployed models. In this paper, we study combination of these two techniques and show that the performance of low-precision networks can be significantly improved by using knowledge distillation techniques. Our approach, Apprentice, achieves state-of-the-art accuracies using ternary precision and 4-bit precision for variants of ResNet architecture on ImageNet dataset. We present three schemes using which one can apply knowledge distillation techniques to various stages of the train-and-deploy pipeline. © Learning Representations, ICLR 2018 - Conference Track *** right reserved.

关键词： Distillation

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：