After reaching the energy efficiency target in 2020 „ AMD 25 × 20 “for mobile processors, the company’s engineers were hungry for a new, daring challenge in energy processor efficiency for the benefit of customers and the well-being of the planet. By 2025, AMD aims to provide 30x increase energy efficiency in the areas of Artificial Intelligence (AI) and Supercomputers (HPC).
![](https://www.itresenja.com/wp-content/uploads/2021/07/21744759_AMD-EPYC-Milan-1P-Server-1024x576.jpeg)
The goal is focused on accelerated computing nodes in data centers, which use AMD processors and AMD GPU accelerators for AI training and HPC applications.
Focused to reduce power consumption while meeting the demands of explosive workloads in the HPC and AI domains, the new 30 × 25 goal will require the company to increase the energy efficiency of the computer node by more than 2.5 times the overall improvements in the industry over the past five years.
Big picture? Incredible benefits from HPC and AI applications with potential savings of billions of kilowatt hours of electricity and potential reduction of computer power consumption by 97% in the next five years.
![](https://www.itresenja.com/wp-content/uploads/2021/09/Screen-Shot-2021-09-29-at-19.59.50-1024x544.jpg)
Background
Looking at current and projected computing requirements, AMD’s team decided it would be most useful to focus to the component that makes up the fastest growing segments in data center preoccupation: accelerated computing.
Accelerated computing nodes are basically a domain-specific configuration that allows for optimal computational performance. At the same time, they enable the most powerful and advanced computer systems in the world. Using these nodes, scientists can train neural networks of artificial intelligence and deliver large-scale supercomputers, to conduct critical research that is at the core of the most promising and much-needed scientific discoveries of our time.
In the next decade, the implementation of AI and HPC will lead to significant developments in various fields, such as climate prediction, genetics, pharmacology, alternative energy, materials science and speech / image recognition…
The enumerated workloads require massive computing power and can therefore be energy intensive. With the growing global demand for the application of artificial intelligence and supercomputers, if companies do not take action, these two areas could dominate the energy consumption in data centers in the coming years.
![](https://www.itresenja.com/wp-content/uploads/2021/09/Screen-Shot-2021-09-29-at-19.59.27-1024x561.jpg)
Why it matters
AMD recognizes the growing but unmet need for continuous, sustainable energy efficiency improvements in accelerated computers. This vision also extends the company’s long-term commitment to environmental management, product energy efficiency and environmental, community and governance (ESG). In order to continue to deliver the computing power the world needs, AMD is inspired by innovation and finding new approaches to raising the energy efficiency curve, in an era of reducing the effects of miniaturization of the semiconductor manufacturing process.
, achieving a 30-fold increase in energy efficiency by 2025 would support the growth of some of the world’s most intensive computing needs, while steps would be taken to mitigate the associated risk of higher energy consumption.
![](https://www.itresenja.com/wp-content/uploads/2021/09/Screen-Shot-2021-09-29-at-19.59.01-1024x550.jpg)
How to get there
To achieve this goal, AMD engineers will give priority achieving a large increase in efficiency in accelerated computing nodes through architecture, processor design, software and innovation in integrated packaging, while publicly measuring progress on an annual basis.
To learn more on a new initiative for energy ef ikasnost, visit: amd.com/en/corporate-responsibility/data-center-sustainability.
Note: This article has been indexed to our site. We do not claim ownership or copyright of any of the content above. To see the article at original source Click Here