Session

DescriptionWith increasing acceleration of computing, data movement has become the major bottleneck in novel/upcoming applications in machine learning, graph analytics, stream processing etc. This session covers novel approaches for accelerating data movement across the design stack ranging from emerging technology to heterogeneous packaging, and domain-specific architectures. In particular, the papers in this session look towards application-driven insights to drive some of the underlying hardware design decisions in memory system design, reconfigurable architectures, photonic interconnections, chiplet-based designs, and 3D ICs.