CC* Compute: GPU Infrastructure to Explore New Algorithmic & AI Methods in Data-Driven Science and Engineering at Tufts University

CC* 计算：探索新算法的 GPU 基础设施

基本信息

批准号：
2018149
负责人：
Christopher Sedore
金额：
$ 40万
依托单位：
Tufts University
依托单位国家：
美国
项目类别：
Standard Grant
财政年份：
2020
资助国家：
美国
起止时间：
2020-07-01 至 2022-06-30
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=2018149&HistoricalAwards=false
关键词：
CC Compute GPU Infrastructure Explore

项目摘要

Advanced scientific and engineering research at Tufts University is employing increasingly complex models, algorithms, simulations and machine learning approaches to large datasets. The larger the dataset or the more complex the model, the longer it takes to compute, slowing down researchers' progress and limiting their ability to innovate. Tufts' addition of six Graphics Processing Unit (GPU) enhanced compute nodes to its high-performance computing cluster accelerates scientific and engineering research in the areas of Algorithm Parallelization & Acceleration and Machine Learning & Deep Learning. Researchers in biology, chemistry, computer science, mathematics, physics, and urban planning leverage the GPU enhanced infrastructure to develop new algorithms and models and accelerate scientific discoveries. Through collaboration with the NSF-funded T-TRIPODS (Transdisciplinary Research in Principles of Data Science) project and the Center for STEM diversity at Tufts, the infrastructure provides new opportunities for underrepresented students to acquire and extend data science and high-performance computing skills.The six GPU enhanced compute nodes are each configured with dual 20-core Intel Xeon Gold 6248 CPUs, 768GB of RAM and 8 NVIDIA Tesla V100 (32GB) GPUs interconnected with NVLink to improve scaling of multi-GPU computation. The nodes are linked with a 100 gigabit network and are accessible to researchers at Tufts and externally through the Open Science Grid (OSG). The large core count, large RAM, interlinked GPU architecture provides the greatest flexibility for researchers to mix traditional and GPU-enhanced approaches in complex computational analysis.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

塔夫茨大学的先进科学和工程研究正在对大型数据集采用日益复杂的模型、算法、模拟和机器学习方法。数据集越大或模型越复杂，计算所需的时间就越长，从而减慢了研究人员的进度并限制了他们的创新能力。塔夫茨大学在其高性能计算集群中添加了六个图形处理单元 (GPU) 增强型计算节点，加速了算法并行化和加速以及机器学习和深度学习领域的科学和工程研究。生物学、化学、计算机科学、数学、物理学和城市规划领域的研究人员利用 GPU 增强的基础设施来开发新的算法和模型并加速科学发现。通过与 NSF 资助的 T-TRIPODS（数据科学原理跨学科研究）项目和塔夫茨 STEM 多样性中心合作，该基础设施为代表性不足的学生提供了获取和扩展数据科学和高性能计算技能的新机会。六个 GPU 增强计算节点均配置双 20 核 Intel Xeon Gold 6248 CPU、768GB RAM 和 8 个 NVIDIA Tesla V100 (32GB) GPU， NVLink 可提高多 GPU 计算的扩展性。这些节点与 100 GB 网络相连，塔夫茨大学的研究人员可以通过开放科学网格 (OSG) 进行外部访问。大核心数、大 RAM、互连 GPU 架构为研究人员在复杂计算分析中混合传统方法和 GPU 增强方法提供了最大的灵活性。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值进行评估，被认为值得支持以及更广泛的影响审查标准。