$AMD CTO Mark Papermaster on Energy Efficiency✍️ As AI usages becomes more and more pervasive across our personal and professional lives, it comes with an insatiable demand for computation. It is an imperative that energy efficiency must scale with it. At AMD, we have been acutely focused on energy efficiency across our product portfolio and particularly for AI. We have achieved an estimated 4x increase in rack-scale AI energy efficiency from 2024 to 2026, ahead of the pace we projected toward our goal of a 20x improvement by 2030. Getting there takes more than advances in any one component. It requires holistic design across silicon, memory, interconnects, software, power and cooling, with the entire system co-optimized for performance per watt. This is increasingly important as cu
在 X 查看 ↗