MRI: track 1: acquisition of a multi-accelerator architecture for advanced scalable computing research

The National Science Foundation Office of Integrative Activities awarded Tennessee Technological University $500,000 on August 1, 2026, under the Integrative Activities program (CFDA 47.083) to acquire a high-performance computing cluster as part of the Major Research Instrumentation (MRI) Track 1 initiative.

The instrument comprises six compute nodes integrating three complementary accelerator technologies—accelerated processing units (APUs), data processing units (DPUs), and field-programmable gate arrays (FPGAs)—operating in concert to enable research on parallel computing architectures and heterogeneous computational systems. The research objective is to advance science and engineering outcomes by demonstrating enhanced performance from multiple heterogeneous computational elements and networks operating together, with applications including real-time inference and collaborative computing workloads.

The instrument will support research and training of undergraduate, master's, and doctoral students at Tennessee Technological University, the University of Tennessee Chattanooga, and the University of Tennessee Knoxville. The award runs through July 31, 2029. Performance takes place in Cookeville, Tennessee.

Abstract

This award funds the acquisition of an advanced computing instrument at tennessee technological university that brings together three complementary classes of accelerator technology in a single, fully networked system. By combining these computing architectures in one machine, the instrument affords researchers a rare opportunity to study how future computers can be designed and programmed to work faster and more efficiently together, and to apply that capability to real scientific and engineering problems. The project advances the national interest by strengthening the computing foundation on which discovery, economic competitiveness, and a skilled technical workforce all depend. The instrument will support the research and training of a group of undergraduate, master's, and doctoral students, broadening participation in advanced computing and providing access to emerging hardware that is not typically available on publicly accessible systems. In doing so, the project promotes the progress of science and contributes to national prosperity and welfare. Science and engineering research to be undertaken includes work from professors and students from tennessee tech, the university of tennessee, chattanooga and the university of tennessee, knoxville. The instrument is a high-performance computing cluster of six compute nodes whose defining feature is the systematic integration of three distinct accelerator technologies operating in concert: accelerated processing units (APUS); data processing units (DPUS); and field-programmable gate arrays (FPGAS). The overarching research theme is to achieve new science and engineering outcomes while demonstrating the enhanced performance attainable from a parallel architecture in which multiple heterogeneous computational elements and networks operate together. The instrument will enable research in real-time inference of collaborative deep neural networks; hyperparameter optimization and neural architecture search; AI-driven blockchain security for the internet of things; parallel programming models spanning multiple forms of computational heterogeneity and hierarchy; smarter network computation; distributed quantum computing simulation and protocols; name-based compute and data placement; AI-driven networking and communications; and flexible, performant collective algorithms generated through a compilation-based method informed by AI/ML. Collectively, this work is expected to inform the design of future cluster and leadership-class computer architectures and to enable new lines of research that depend on the close coordination of diverse accelerators. 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.- subawards are not planned for this award.