Multi-Die Design for HPC Applications

High-performance computing (HPC) has evolved dramatically over the years to address the evolving complex requirements of today’s workloads. Initially, the term referenced specialized supercomputers to solve complex scientific and engineering problems. In addition to AI, in recent years, HPC applications have increased to include:

• Automotive including advanced driver-assistance systems (ADAS) and in-vehicle infotainment systems (IVI)
• Data center networking
• Cloud data center applications
• Many others, from medical imaging, to genomics, to weather forecasting 

This white paper explains how the semiconductor ecosystem is adopting multi-die designs to address the performance, power, area, and latency requirements of HPC chips, which are on the path to trillion-transistor designs.