Principled Algorithm, Data Structure and System Design for Processing-in-Memory

Explore principled approaches to algorithm, data structure, and system design specifically tailored for Processing-in-Memory (PIM) architectures in this 26-minute conference talk. Learn how to develop computational methods that leverage the unique capabilities of PIM systems, where processing units are integrated directly with memory components to reduce data movement overhead. Discover the fundamental design principles that guide the creation of efficient algorithms and data structures optimized for PIM environments, and understand the system-level considerations necessary for implementing these solutions effectively. Gain insights into the challenges and opportunities presented by PIM technology, including how to rethink traditional computational approaches to take advantage of near-data processing capabilities and minimize the memory wall bottleneck that affects conventional computing architectures.