Texas A&M University College of Engineering faculty Dr. Dilma Da Silva and Dr. Raymundo Arroyave are co-principal investigators on a research project that has received a $3.9 million Major Research Instrumentation grant from the National Science Foundation (NSF). The grant will allow the university to acquire a next-generation, composable high-performance computing platform to enable researchers to make transformative advances in a wide range of scientific fields.
“Acquiring such hardware capabilities will enable Texas A&M scientists and engineers to explore new ways of leveraging machine learning in data-driven discovery” said Da Silva, computer science and engineering professor and holder of the Ford Motor Company Design Professorship II. “For computer researchers like me, understanding the demands that large-scale data analytics make on the platform may point out novel directions to accelerate the software.”
With the support of the grant, Texas A&M’s High Performance Research Computing (HPRC) will purchase FASTER (Fostering Accelerated Scientific Transformations, Education and Research), a composable high-performance data-analysis and computing instrument. FASTER will significantly benefit scientific fields that rely on artificial intelligence and machine learning (AI/ML) techniques, big-data practices and high-performance computing (HPC) technologies.
These fields include the development of AI/ML models, cybersecurity, health population informatics, genomics, bioinformatics, computer-aided drug design, agricultural sciences, life sciences, biophysics, oil and gas simulations, materials science, climate modeling, multi-scale simulations, quantum computing architectures, biomedical imaging, geosciences and quantum chemistry.
“This is an important addition to our already impressive capabilities in high-performance research computing,” said Dr. Mark A. Barteau, vice president for research. “Computational science has become the ‘third pillar’ of research and scientific investigation, and is essential for advanced theory and experimentation. In this age of groundbreaking, multidisciplinary research, it is vital for a world-class institution like Texas A&M to offer researchers access to the paradigm-changing power of high-speed computation and data analysis.”
Texas A&M will provide cost sharing of more than $1.32 million toward the project, which will be used to support researchers to effectively use this novel HPC/AL/ML platform.
“Congratulations to HPRC and the project team for securing the federal funding for this vital addition to our computing capabilities,” said Dr. Costas N. Georghiades, senior associate vice president for research. “Research workflows in an ever-growing number of scientific and engineering disciplines are becoming more and more reliant on high-performance computing to pursue pioneering discoveries and innovations. We are proud to expand and extend this capacity to our outstanding researchers across the A&M System.”
In addition, 30% of FASTER’s computing resources will be allocated to researchers nationwide by the NSF’s Extreme Science and Engineering Discovery Environment program. The platform also will contribute to code development, education and the workforce development goals of the NSF advanced cyberinfrastructure ecosystem.
“This unique HPC/AI/ML platform delivers much desired composability-features that go beyond the scope of the current generation of deployed supercomputers,” said Dr. Honggao Liu, HPRC director and principal investigator for the FASTER project. “The FASTER platform will help researchers nationwide seek answers to questions that are currently intractable.
“The FASTER platform removes significant bottlenecks in research computing by leveraging a technology that can dynamically allocate resources to support research workflows. FASTER combines the innovative composable software-hardware approach with cutting-edge technologies such as next-generation central processing units and graphic processing units (GPUs), state-of-the-art non-volatile memory express (NVMe)-based storage and high-speed interconnect. Workflows on FASTER will dynamically integrate GPUs and NVMe to compose a single node, allowing them to scale beyond traditional hardware limits. This will allow researchers to use resources far more efficiently and to conduct more research in less time.”
Arroyave is a professor in the Department of Materials Science and Engineering. Dr. Zhe “Sarina” Zhang is an assistant professor in the Department of Geography at Texas A&M and co-principal investigator on the project.