CC* Compute: OneOklahoma Cyberinfrastructure Initiative Research Accelerator for Machine Learning (OneOCII-RAML)

点击翻译按钮获取中文摘要

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).In this project, the University of Oklahoma (OU) aims to acquire a GPU powered Artificial Intelligence/Machine Learning HPC resource, “OneOklahoma Cyberinfrastructure Initiative Research Accelerator for Machine Learning (OneOCII-RAML)”. OneOCII-RAML is a 7 node cluster, with each node configured with an NVIDIA HGX A100 4-GPU (28 A100 GPUs total). The system will directly support several active research projects, educational programming to grow the use of AI/ML in research beyond Computer Science disciplines, and would be a state-wide resource via the OneOklahoma Friction Free Network (OFFN). OneOCII-RAML would be integrated with an in-progress supercomputer deployment at OU. OneOCII-RAML's role is to accelerate Oklahoma's strategic research in many STEM disciplines, including: trustworthy AI in weather, climate and coastal oceanography; the ATLAS high energy physics collaboration; smoking cessation; bioinformatics and medical physics imaging; human brain signals; mobile network design; reservoir rock typing; natural language processing; and quasar feedback.This University of Oklahoma (OU) project focuses on acquiring, deploying and maintaining a large scale Artificial Intelligence/Machine Learning (AI/ML) computer system -- the OneOklahoma Cyberinfrastructure Initiative Research Accelerator for Machine Learning (OneOCII-RAML). The computer system will enable researchers (undergraduates, graduate students, postdocs, staff, faculty) at OU and non-commercial institutions across Oklahoma to pursue AI/ML-based research and education via Graphics Processing Units (GPUs). OneOCII-RAML's hardware platform is a set of GPU server computers with NVIDIA Ampere Next GPUs, integrated into OSCER's new supercomputer. The project will also teach researchers how to use GPUs for AI/ML. OSCER's Research Computing Facilitators will be the mechanism for working one-on-one with each research team on how to use these GPUs. Eighteen research teams have been identified that are already using GPUs for AI/ML and fifty-one research teams aspire to do so. A rapidly growing portfolio of transformative Science, Technology, Engineering & Mathematics (STEM) research is based on AI/ML calculations, which increasingly depend on accelerators such as GPUs for reducing time-to-science. The project makes a strong case for science-driven needs for this GPU-based addition to OU compute resources with high probability of use and impact throughout the state.This project is funded through the collaborative efforts of the Office of Advanced Cyberinfrastructure (OAC) and the Established Program to Stimulate Competitive Research (EPSCoR).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.