Another week, another Rambus Design Summit wrap up! Over the coming weeks, we will be featuring sessions from RDS 2021 with key takeaways and speaker highlights. This week we are featuring…
Session Topic: Emerging Compute Architectures for the Evolving Data Center
Rambus fellow and distinguished inventor, Dr. Steven Woo, kicked off the Rambus Design Summit with some insights about compute architectures and the future of the data center.
In his talk, he discussed that with the exponential growth of digital data – doubling every two to three years – all processing architectures, whether in the data center, at the edge or in end points, have evolved to become more data centric. However, the industry will face some key challenges in meeting the need to improve performance and power efficiency, without tools like Moore’s Law and Dennard Scaling.
He provided his insight into the key drivers that will shape the evolution of the data center, including:
- The move to disaggregation and composability
- AI processing/AI training
“Data centers are beginning to adopt many of the things that we’ve seen in the high performance computing world. And we’re generally seeing a greater convergence between both the technology and the application approaches in both the high-performance computing world and within data centers, and we expect this to continue moving forward.”
Though there will be physical challenges in improving memory and interconnect performance and power efficiency, he mentions that it is an exciting time to be in the industry. New memory technology, such as DDR5 and CXL, will help enable technology towards the goal of a disaggregated future. In addition, AI and other demanding applications, will continue to influence the way these architectures continue to evolve.
Check out Steve’s full talk here.
About Steven Woo
Steven Woo is a Fellow and Distinguished Inventor at Rambus Inc. Since joining Rambus, Steve has worked in various roles leading architecture, technology, and performance analysis efforts, and in marketing and product planning roles leading strategy and customer programs. Steve has over 30 patents in the areas of memory, memory systems, and neural networks. Steve received his PhD and MS degrees in Electrical Engineering from Stanford University, and Master of Engineering and BS Engineering degrees from Harvey Mudd College.