Director: Tajana Rosing
Assistant Director: Nam Sung Kim
PRISM’s goal is to explore revolutionary pathways that will deliver long-term industry impact. We will build adaptive and flexible HW or SW “prisms”, to replace the “blocks” that are a byproduct of today's rigid abstraction layers. Combinations of such HW and SW components will expose the right capabilities and interfaces at the right time and place, resulting in holistic cross-layer IMS codesign, without affecting programming complexity. Instead of a rigid hierarchy that exists today, we envision an IMS system where the software layer adapts to existing frameworks for compatibility and distributes work throughout the system for maximum performance and efficiency while ensuring security. The architecture layer unifies near-data and distributed computing by leveraging CXL and heterogeneous chiplet integration technologies, efficiently and securely integrates traditional and computational memory sub-systems. To accomplish these goals, the center is divided into four research themes:
- Theme 1: Systems & Software
- Theme 2: Architecture
- Theme 3: Devices & Circuits
- Theme 4: Grand Challenges
Grand challenge applications: We will demonstrate our prototype system in the context of two applications of high priority for societal, commercial, and national security: personalized medicine and deep analytics. Both have a huge demand for bandwidth and capacity, while requiring accurate and timely results. For deep analytics, our goal is to enable 100x more insights to be gleaned within a given capacity and bandwidth compared to the state of the art. This will be realized through cross-stack innovations that change how, where and when data is stored, accessed, processed, or moved. We plan to develop new benchmarks and datasets by leveraging existing public domain data (via the IEEE DataPort) and workloads from our collaborators. Our work will not only revolutionize the drug discovery and data analytics, but will also develop and test systems needed to accelerate other data-intensive workloads, as these applications share similar constituents, such as databases, sparse data processing, algorithms based on graphs, and machine learning algorithms, such as clustering.