High Performance Computing and Machine Learning Infrastructure for Oregon Life Sciences

俄勒冈生命科学的高性能计算和机器学习基础设施

• 批准号: 10630777
• 负责人: Kyle Ellrott
• 金额: $ 195.5万
• 依托单位: OREGON HEALTH & SCIENCE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-21 至 2024-09-20
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10630777
• 关键词: Address; American; Area; Artificial Intelligence; Biological Sciences; Collaborations; Communicable Diseases; Communities; Comprehensive Cancer Center; Computational Biology; Data; Dementia; Diabetes Mellitus; Equipment; Funding; Genomics; Grant; Growth; Health; High Performance Computing; Infrastructure; Institution; Legal patent; Machine Learning; Malignant Neoplasms; Memory; Minor; Modeling; Oregon; Research; Research Personnel; Research Project Grants; Resources; Science; Services; System; United States National Institutes of Health; Universities; cluster computing; computational platform; computing resources; end of life; improved; innovation; mathematical model; open data; radiomics; tool

Project Summary/Abstract Oregon Health & Science University (OHSU) requests funds for the “High Performance Computing and Machine Learning Infrastructure for Oregon Life Sciences” to refresh outdated equipment and expand existing hardware resulting in a 39% net increase in memory and 183% increase in GPUs. Exacloud, the current high- performance computing (HPC) infrastructure at OHSU and Knight Cancer Institute was first installed in 2014 and expanded in subsequent years. The improvements will meet current demands, support recent NIH-funded growth in quantitative sciences, leverage the shared expertise of computational researchers at OHSU, and meet institutional needs at least until 2026. This proposal serves 61 life sciences research groups in the Portland area, 43 of which are actively funded by NIH and 26 are presented here as Major or Minor users who are listed as Contact PI on NIH grants totaling more than $33 million in annual funding. The scientific use cases for this computational platform include cancer, diabetes, infectious disease, genomics, radiomics, artificial intelligence, and modelling research. Overall, NIH research projects with over $89 million in annual NIH funding are listed in this proposal which would directly benefit from the improved system. The current OHSU Exacloud operates a 282-node, 9176-core high-performance computing (HPC) cluster. 36% of the current hardware is end-of-life, slowing OHSU research. We seek funds to purchase 102 nodes with 768 GB memory each (3264 cores and 176 GPUs). After discontinuing 127 nodes we will leverage the existing Exacloud infrastructure to create a 257 node/8888 core cluster with 282 GPUs that can accommodate 15% annual growth in computing and data intensive science for OHSU users as the new CPU cores are 86% faster than the equipment they will replace. This resource will enable open science, support cross-institutional collaboration, and manage increased demands on computational resources leveraging shared OHSU / Knight research needs and OHSU shared services computational expertise. OHSU and the Knight Cancer Institute and achieve more than $330 million of NIH funding per year. The Knight Cancer Institute is the only Oregon-based NCI-designated Comprehensive Cancer Center with innovative research, including Computational Biology and Machine Learning. OHSU performs high-impact global research as a preeminent research university in Oregon and is 12th among American research universities in generating new patents. OHSU investigators use innovative tools such as machine learning, genomics, and mathematical modeling to address important high-impact health concerns including cancer, infectious diseases, diabetes, and dementia. The improved Exacloud will support this vibrant life science research community and provide a platform for health-focused innovations.

项目摘要/摘要 俄勒冈州健康与科学大学（OHSU）请求资金进行“高性能计算和 俄勒冈生命科学的机器学习基础设施”以刷新设备并扩展现有 硬件导致记忆净增加39％，GPU增加了183％。 Exacloud，当前高 OHSU和Knight Cancer Institute的性能计算（HPC）基础设施于2014年首次安装 并在随后的几年中扩展。这些改进将满足当前需求，支持最近的NIH资助 定量科学的增长，利用OHSU计算研究人员的共同专业知识，以及 至少在2026年之前满足机构需求。该提案为61个生命科学研究小组提供 波特兰地区，其中43个由NIH积极资助，这里有26个是主要用户或次要用户 列为NIH赠款的Contact PI，年度资金总额超过3,300万美元。科学用途 该计算平台的病例包括癌症，糖尿病，传染病，基因组学，放射素学， 人工智能和建模研究。总体而言，NIH研究项目年度超过8900万美元 该提案中列出了NIH资金，该提案将直接受益于改进的系统。 当前的OHSU Exacloud运行一个282节点，9176核高性能计算（HPC）群集。 36％ 当前的硬件是生命终止，从而减慢了OHSU研究。我们寻求资金以768的价格购买102个节点 GB内存每个（3264个内核和176 GPU）。停止127个节点后，我们将利用现有的 Exacloud基础架构可创建257节点/8888核心群集，并具有282 GPU，可容纳15％ 由于新的CPU核心的年增长 比他们将更换的设备。该资源将使开放科学能够支持跨机构 协作并管理利用共享OHSU /骑士的计算资源的需求增加 研究需求和OHSU共享服务计算专业知识。 OHSU和骑士癌症研究所，每年获得超过3.3亿美元的NIH资金。骑士 癌症研究所是唯一具有创新性的基于俄勒冈州的NCI指定的综合癌症中心 研究，包括计算生物学和机器学习。 OHSU进行高影响全球研究 作为俄勒冈州著名的研究大学，在美国研究大学中排名第12 新专利。 OHSU调查人员使用创新工具，例如机器学习，基因组学和数学 建模以解决重要的高影响健康问题，包括癌症，传染病，糖尿病， 和痴呆症。改进的Exacloud将支持这个充满活力的生活科学研究社区，并提供 以健康为中心的创新平台。