Keynote: The Road to 1 Trillion Begins at the Edge
summary
Claims a path to one trillion intelligent devices by 2035 and a $1T IoT market, argues that infrastructure must be highly efficient and heterogeneous, and says ARM delivers ~3x compute density by 2020 via workload-optimized SoCs. Demonstrations include project Trillium IP for on-device ML, an ONF/OPNFV 5G edge demo using ARM+GPU showing a 1U prototype at 4.4x performance, 30% power, 40% cost, and tiny 2–4W microservers handling facial-recognition slices without host PCI hops. Tools: Kubernetes/Helm, OpenStack, CI/CD, DPDK.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:30The Big Picture
- 2:28Project Trillium
- 3:44Efficiency
- 8:59Linux Foundation Network
- 10:16Use Cases
- 11:25Reference Design
- 12:00Telco Systems
- 13:00Mobile World Congress
- 14:09MicroServer Demo
- 16:58OPNFE
- 18:13Conclusion
related talks
21mA New Architecture for the Intelligent Edge
01h 16mOpening Keynote
22mAdvancing the Human Experience
14mKeynote: RISC-V Opportunities at the Edge of AI
26mKeynote: Artificial Innovation Accelerates Artificial Intelligence
58mThe Next Decade of Software Development
20mHow ML is Powering the Future of IoT
29mEdge Intelligence Revolution