Processing with Intelligent Storage and Memory – PRISM Center
Data volumes are growing faster than Moore’s Law and are outstripping the total capacity of silicon production. Future personalized, secure, data-intensive, and interactive applications require massive and distributed memory and storage. Systems have to compute in real time – they need to digest the data as it is being generated, integrate it, select carefully what should be stored and where, all the while optimizing for energy efficiency, reliability and security. Today’s systems are bottlenecked by data movement costs.
To achieve better performance, PRISM systems will keep the data in place as much as possible, and move computing close to the data. This requires re-architecting the overall system, from the virtualization layers, where we need to account for highly distributed and disaggregated data, to the development of novel devices that can scale beyond what is possible today, while allowing for in-situ computing. PRISM will blur rigid boundaries between hardware and software to enable flexible cross-layer co-optimizations while making these changes transparent to the application designer. The systems will gracefully scale out and down depending on the needs of the applications and capabilities of the hardware infrastructure. By radically re-architecting memory and storage to enable new computing functionalities for improved energy efficiency and performance, PRISM will achieve near “speed-of-light computing,” revolutionize computing systems across multiple application platforms, including high-performance and embedded computing, and provide quantum leaps in capability, scale, and QoS in PRISM’s grand challenge applications. Changes are expected to permeate all layers of the stack from the application layer, through the operating system, and the underlying system and memory subsystem architectures, including the various abstractions along the data path. Holistic system consideration will encompass necessary enhancing components such as hypervisors and virtual memory systems, as well as data layout, security, and management. PRISM will demonstrate at least 100x improvement in performance and efficiency in two classes of grand challenge applications: “Personalized and Secure Drug Development”, and “Deep Insights.”
PRISM received a $35 million grant from the Semiconductor Research Corporation. The grant is part of the JUMP 2.0 program, which seeks to significantly improve performance, efficiency, and capabilities across a range of electronics systems. Novel materials, devices, architectures, algorithms, designs, integration techniques, and other innovations are at the heart of problem-solving for next-generation information and communications challenges. JUMP 2.0 brings together SRC, DARPA and industry and academic stakeholders.
Additionally, a group of 10 universities banded together to contribute additional funds to create the $50.5 million UC San Diego-led Processing with Intelligent Storage and Memory center, or PRISM. Partnering institutions in PRISM include Stanford University; Georgia Tech; University of California at Los Angeles, Berkeley and Irvine; University of Wisconsin Madison; Penn State University; University of Illinois Urbana-Champaign; University of Texas at Austin; and University of Virginia.