With more scientific fields relying on neural networks (NNs) to process data incoming at extreme throughputs and latencies, it is crucial to develop NNs with all their parameters stored on-chip. In many of these applications, there is not enough time to go off-chip and retrieve weights. Even more so, off-chip memory such as DRAM does not have the bandwidth required to process these NNs as fast as the data is being produced (e.g., every 25 ns). As such, these extreme latency and bandwidth requirements have architectural implications for the hardware intended to run these NNs: 1) all NN parameters must fit on-chip, and 2) codesigning custom/reconfigurable logic is often required to meet these latency and bandwidth constraints. In our work, we show that many scientific NN applications must run fully on chip, in the extreme case requiring a custom chip to meet such stringent constraints.
Rethink the Role of Neural Decoders in Quantum Error Correction
College of Computing;Data Science, Nanyang Technological University, Singapore 639798, Singapore;Department of Electrical Engineering;Computer Science, Tokyo University of Agriculture \& Technology, Koganei, Tokyo, 184-8588, Japan;School of Physical;Mathematical Sciences, Nanyang Technological University, Singapore 639798, Singapore
Benchmarking energy consumption and latency for neuromorphic computing in condensed matter and particle physics
University of Twente(特文特大学);Radboud University(拉德堡德大学);IBM Research Europe - Zürich(IBM欧洲研究院-苏黎世);Nikhef(荷兰国家亚原子物理研究所);Eidgenössische Technische Hochschule Zürich(苏黎世联邦理工学院);SURF Cooperation(SURF合作组织);University of Valencia-CSIC(瓦伦西亚大学-西班牙国家研究委员会)
Hardware Co-Design Scaling Laws via Roofline Modelling for On-Device LLMs
AI Lab, The Yangtze River Delta Institution of Automation, Chinese Academy of Sciences(人工智能实验室,长江三角洲自动化研究所,中国科学院);University of Chinese Academy of Sciences(中国科学院大学);University College London(伦敦大学学院);Institution of Automation, Chinese Academy of Sciences(自动化研究所,中国科学院);The University of Edinburgh(爱丁堡大学)
Reliable edge machine learning hardware for scientific applications
Fermilab(费米实验室);University of Pisa(比萨大学);UC San Diego(加利福尼亚大学圣迭戈分校);UC Berkeley(加利福尼亚大学伯克利分校);LBNL(劳伦斯伯克利国家实验室);ICSI(国际计算机科学研究所);MIT(麻省理工学院);Drexel University(德雷塞尔大学)
Inferring Convolutional Neural Networks' accuracies from their architectural characterizations
Rhodes College(罗德学院);The University of Iowa(爱荷华大学);Fermi National Accelerator Laboratory (FNAL)(费米国家加速器实验室(FNAL));Oak Ridge National Laboratory(橡树岭国家实验室);Universidad Técnica Federico Santa María(费德里科·圣玛丽亚技术大学)
Effectiveness and predictability of in-network storage cache for scientific workflows
University of California at Berkeley(加州大学伯克利分校);Lawrence Berkeley National Laboratory(劳伦斯伯克利国家实验室);Energy Sciences Network(能源科学网络);University of California at San Diego(加州大学圣迭戈分校);California Institute of Technology(加州理工学院)