Moore’s law, named after the American chemist and co-founder of Intel Corporation, is the observation that the number of transistors on microprocessor integrated circuits double every two years. Although the trend is expected to slow down soon, it has been consistently observed in the computer sector since the 1970s. In practice, the steady growth in the number of transistors has meant an increase in computing power, as well as – until recently – higher electricity consumption. The combination of rising energy prices and the fast-growing mobile devices sector makes curtailing energy use from appliances equipped with microprocessors (computers, tablet computers and smart phones) a business as well as environmental challenge.
How energy efficient are today’s microprocessors?
In order to assess the extent of the progress made in terms of energy efficiency, it is useful to distinguish an appliance’s energy use from its efficiency. Whereas the former is expressed in Kwh, the latter is a performance-to-KwH-used ratio. Besides, a device’s component parts must be separately looked at, as some of them represent significant potential energy savings.
Taking a practical example, a 2009 study by Intel confirmed Moore’s law’s relevance to CPUs (central processing units i.e. microprocessors), although IBM came to slightly different results, i.e. that the volume of digital data processing has gone up by a factor of 69 in 10 years. In any case, the sector’s key players agree that the computing power of CPUs increased exponentially, which in turn affects the amount of energy they need to operate.
Steadily rising electricity use of Intel processors since the 1950s
Source: ASSESSING TRENDS IN THE ELECTRICAL EFFICIENCY OF COMPUTATION OVER TIME, 2009 study by Intel
Compared to a computer’s other components, processors remain therefore particularly energy-intensive. Their energy consumption depends on their tension and clock rate (DVFS); it is also often expressed in terms of thermal design power (TDP), i.e. the maximum amount of power the cooler is required to dissipate. However, TDP provides only one part of the picture, as power management capacity must also be taken into account. Today the third version of APCI (advanced configuration and power interface) specifications provides a standard for device configuration and fine tuning of power management according to each component’s specific energy needs.
Microprocessor energy efficiency at the heart of technological arena
Energy use currently accounts for up to 40% of operating costs in a data centre (facilities consisting of a very large number of processors). Following the boom of cloud computing (i.e. storing data on data-centre-based networks instead of local hard drives), EU authorities launched the Eurocloud project aimed at cutting servers’ energy consumption in data centres by a factor of 10. Furthermore, the EU invested 3.3 billion € supplemented by 40% private funding in order to develop a new processing technology, so-called “multi-core” processors.
Although several European companies and research centres are involved in this initiative (ARM, Nokia, IMEC, EPFL, Cyprus University), the American Intel corporation wishes to maintain its leadership position and is working on a new generation of processors with smaller TDP (from 40 to 17W) thanks to finer wafer etching. AMD, another computer giant, decided to improve processor design itself and now offers “hybrid” chips (APU: accelerated processing unit) adding an extra unit next to the main one, responsible for specific computing operations (resolution, peripheral control…). Finally, last December in Paris, the British ARM showcased a processor featuring new system architecture, scalable performance and low power consumption.
For more information :