AMD today announced its ambitious goal of achieving thirty times the energy efficiency of its EPYC CPUs and Instinct GPU accelerators by 2025. AMD itself, which is a very ambitious goal and in a very short time, since it would exceed efficiency improvements of the entire industry by 150%.
This energy efficiency goal from AMD comes from a huge increase in processing requirements for accelerated compute nodes for high-level functions such as AI training, weather prediction, genomics, and supercomputer simulations to big scale. If AMD reached its goal, the total power consumption of these systems would be reduced by 97% over a five-year period.
AMD advocates for lower power CPUs
According to AMD's Mark Papermaster, achieving gains in processor power efficiency is a long-term design priority for them and they are now setting a new goal for modern computing nodes when applied to AI training and high-performance computing deployments. . They will focus on these segments so that leading companies improve their environmental management.
AMD has already been exploring a host of energy-efficiency improvements in its CPU and GPU designs, with AMD's Zen CPUs outperforming Intel's in performance-per-watt ratio. It also made power consumption improvements to its RDNA 2 GPUs. Part of these improvements can be attributed to the node hopping fabrication, at least on the GPU side.
The manufacturer predicts that technologies such as 3D cache stacking and an increasingly efficient engineering approach will be required. Exact techniques to achieve these goals have not been announced, but if true, the energy savings will make many users concerned about the environmental impact or monetary outlay want to bet on AMD CPUs in order to consume less electricity.
Source: Tom's Hardware
