Project Grant 2615713

The National Science Foundation Division of Information and Intelligent Systems awarded Boise State University $499,837 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop transformative AI hardware that integrates photonic and electronic computing chips near memory through vertical 3D assembly, reducing data transfer costs and energy consumption for artificial intelligence computing.

The project, titled "3D-PFLOPS: 3D Integrated Parallel Fabrics Enabling Layered Opto-Electronic Processors for Near-Memory AI Computing," creates an architectural simulation framework for phase-aware heterogeneous processors that combine electronic and photonic processing units integrated near shared memory. The work addresses a critical bottleneck in current AI hardware: the energy and overhead costs of moving large amounts of data between processing and memory chips. By assigning different AI computing phases—training and inference—to the chip best suited for each task, the design optimizes both computational and data communication costs compared to existing technologies.

The three specific research objectives are: co-designing an electro-optical application-specific integrated circuit; developing a thermal-aware high-bandwidth near-memory 3D integration strategy; and creating a topology-aware, localized data orchestration and task-scheduling framework that enables efficient execution across heterogeneous compute fabrics. Performance occurs in Boise, Idaho, with a period of performance from August 1, 2026, through July 31, 2029. The assistance type is a Project Grant. Boise State University is the prime recipient and the ultimate recipient.