Cybersecurity for Future Extreme Computing
summary
High-performance computing is a specialized, shared system that uses exotic interconnects and software, serves hundreds of users who resist overhead, and faces real risks from globalization and aging designs. The talk claims the Top500 LINPACK race overvalues FLOPS while real workloads lag, notes China now holds ~214 Top500 slots, predicts exascale machines cost $20–$400M and draw 20–65 megawatts, cites Moore's Law limits and global chip supply risks, and urges design for data, onshoring fabs, and project-based cybersecurity hires.
AI-written summary · about
chapters (10)
- 0:00Cybersecurity for extreme computing
- 2:31HPC's critical role
- 5:21Defining high-performance computing and its history
- 13:11HPC globalization challenges
- 18:33HPC community culture
- 22:10HPC decline and Top500 limits
- 25:40Moore's Law end and Big Tech rise
- 29:22Extreme-scale computing risks
- 33:38HPC security advantages
- 35:28Policy for secure HPC
related talks
38mEvery Chip Counts: Decentralized Cloud Rebellion
57mKeynote: Information Technology Systems in a Post-Silcon World
52mDefense At Hyperscale: Technologies and Policies for a Defensible Cyberspace
14mThe Future of Cybersecurity in the Age of Exponential Change
43mOutpacing the Pacing Challenge: Cyber Command's Strategic Scaling
13mThe Future of Cyber Security pt.1
01h 04mThe future & challenges of cloud
42mSponsored Session: Confidential Computing - New Capabilities for New Workloads