Scaling and Performance Challenges of Future DRAM
Shigeru Shiratake
Abstract
Over the years, memory and storage performance requirements have been driven by growth in computing and data needs. The future brings continued demands for performance improvements in terms of speed and power, as artificial intelligence, high-performance computing, and data analysis continue to become integrated into every aspect of modern life. Artificial Intelligence has become a game-changer—harvesting enormous data sets and converting them to insights and intelligent action in automotive, mobile edge devices, enterprise storage, and medical imaging applications. Historically, DRAM memory technology has scaled to achieve enough cost per bit reduction and meet high-speed with low-power requirements. Also, high volume manufacturing has been transformed with ten nanometer-class process technology. As future performance demands continue to increase, however, DRAM will run into many challenges to achieve cell devices below 1znm technology. It will take innovative processes and material development to overcome these challenges and provide the breakthroughs and innovative design approaches needed to deliver on future performance demands and scaling requirements.