If AMD gets its wish, the company's APUs will be 25 times more efficient by the year 2020. In a keynote presentation at the China International Software and Service Fair today, AMD CTO Mark Papermaster laid out both the strategy and the practical steps the company intends to take to get there.
AMD claims to be taking a three-pronged approach toward improving efficiency. First is further development of the company's Heterogeneous Systems Architecture (HSA). Optimally, HSA can help a system take better advantage of its computing resources and automatically choose the most appropriate resource to complete a given task. Both of these improvements should reduce power consumption. Second up is more fine-grained active power management, wherein an APU monitors the types of tasks it's running and adjusts clock speed and voltage accordingly. Finally, AMD claims to have a number of proprietary, hardware-level innovations in the pipeline that will further reduce power consumption, including "inter-frame power gating, per-part adaptive voltage, voltage islands, [and] further integration of system components."
According to research by Dr. Jonathan Koomey, which AMD cites in its announcement, improvements in power efficiency tend to track closely with Moore's Law. However, the company claims that it can outpace this "natural" improvement in efficiency by 70% between 2014 and 2020 by applying the above innovations.
How did AMD get its efficiency figures? The company says it arrived at these measurements "by taking the ratio of compute capability as measured by common performance measures such as SpecIntRate, PassMark and PCMark, divided by typical energy use as defined by metrics such as ETEC (Typical Energy Consumption for notebook computers) as specified in Energy Star Program Requirements Rev 6.0 10/2013."
Based on our experiences with the company's Kaveri APUs, the company still has some work to do on the efficiency front. Let's hope these improvements will help it become more competitive over the next few years.