Energy Efficiency (2016) 9:321–338 DOI 10.1007/s12053-015-9371-1 ORIGINAL ARTICLE Taming the energy use of gaming computers Nathaniel Mills & Evan Mills Received: 11 December 2014 /Accepted: 8 June 2015 /Published online: 20 June 2015 # Springer Science+Business Media Dordrecht (outside the USA) 2015 Abstract One billion people around the world engage in scoping estimate suggests that gaming PCs consumed some form of digital gaming. Gaming is the most energy- 75 TWh/year ($10 billion) of electricity globally in intensive use of personal computers, and the high- 2012 or approximately 20 % of total PC, notebook, and performance Bracecar^ systems built expressly for gam- console energy usage. Based on projected changes in the ing are the fastest growing type of gaming platform. installed base, we estimatethatconsumptionwillmore Large performance-normalized variations in nameplate than double by the year 2020 if the current rate of power ratings for gaming computer components available equipment sales is unabated and efficiencies are not on today’smarketindicatesignificantpotentialforenergy improved. Although they will represent only 10 % of savings: central processing units vary by 4.3-fold, the installed base of gaming platforms in 2020, relatively graphics processing units 5.8-fold, power supply units high unit energy consumption and high hours of use will 1.3-fold, motherboards 5.0-fold, and random access result in gaming computers being responsible for 40 % of memory (RAM) 139.2-fold. Measured performance of gaming energy use. Savings of more than 75 % can be displays varies by 11.5-fold. However, underlying the achieved via premium efficiency components applied at importance of empirical data, we find that measured peak the time of manufacture or via retrofit, while improving power requirements are considerably lower than name- reliability and performance (nearly a doubling of perfor- plate for most components tested, and by about 50 % for mance per unit of energy). This corresponds to a potential complete systems. Based on actual measurements of five savings of approximately 120 TWh/year or $18 billion/ gaming PCs with progressively more efficient component year globally by 2020. A consumer decision-making configurations, we estimate the typical gaming computer environment largely devoid of energy information and (including display) to use approximately 1400 kWh/year, incentives suggests a need for targeted energy efficiency which is equivalent to the energy use of ten game con- programs and policies in capturing these benefits. soles, six standard PCs, or three refrigerators. The more intensive user segments could easily consume double this Keywords Information technologies . Computing central estimate. While gaming PCs represent only 2.5 % energy use . Gaming computers of the global installed PC equipment base, our initial Context N. Mills http://GreeningTheBeast.org In the quest for technological performance improvements, the racecar is often invoked as a locus of innovation. In the E. Mills (*) Lawrence Berkeley National Laboratory, Berkeley, USA energy sector, this analogy has been applied to data cen- e-mail: [email protected] ters as energy-intensive environments where significant 322 Energy Efficiency (2016) 9:321–338 Fig 1 A surround setup representing the epitome of desktop aggressively tiered electric tariffs), comparable with a highly effi- gaming. A system such as this could approach 2000 W of name- cient home. The underlying machine possesses two 500-W AMD plate power, including displays and peripherals. Based on actual R9 295X2 graphics cards and a 1500-W power supply unit. measured demand, used 8 h/day in gaming mode, the system Sources: HardwareCanucks (2014)andhttps://twitter.com/ would consume roughly 3500 kWh/year (perhaps $1400 with elmnator innovations have been made in IT equipment as well as Alliance 2013). A small subset of people use their the surrounding heating, cooling, and power-delivery in- computers exclusively for gaming, while most engage frastructure (Mills et al. 2007). Similarly, at the distributed in the typical array of computer activities. Even game scales of personal computing, the high-performance gam- consoles have become general media devices. Game ing computer (we subsequently refer to these by the consoles (e.g., PlayStation, Nintendo, and Xbox) have shorthand Bgaming computers^)(Fig.1)hasbeenthe received most of the attention within the energy com- focus of efforts to boost performance in order to meet munity, often to the exclusion of far more energy inten- rapidly increasing user expectations (Short 2013). sive gaming computers (Urban et al. 2014). There are Estimates placed the flow of digital media to US wide variations and strong trends in the choice of plat- households at 6.9 zettabytes (ZB; 1021 bytes) per forms, with the installed base of game consoles year in 2012, of which 2.5 ZB (34 %) was attributed projected to decline and that of desktop gaming com- to gaming (Short 2013). US households are puters to increase (Fig. 2). projected to spend 211 billion hours of gaming in The global count of people utilizing gaming com- 2015, more than the time spent on the telephone, puters was estimated at 54 million in 2012 (33 countries mobile computing, or messaging. Use has doubled studied) and projected to grow to 72 million together since 2008. The 43.6 million Bextreme^ and Bavid^ with sales of related computer hardware of $32 billion gamers spend 4.4 h/day in the activity (all platform by 2015 (Business Wire 2012). About half of the 100 types) versus 7.2 h/day for the 10 million Bextreme^ million PCs with discrete graphical processing units gamer subgroup (Short 2013). (GPUs) shipped in 2014 were purchased by consumers, An estimated one billion people globally engage in with the other half destined for workplace environments some form of personal computer gaming (PC Gaming (Peddie 2014). Fig. 2 BEnthusiast^ gaming computers are a small but growing segment of gaming platforms (a rough proxy for the aforementioned BExtreme^ and BAvid^ user groups), with consoles projected to decline. This chart shows the installed base (stock), with projections from 2014. Excludes mobile platforms (adapted from Open Gaming Alliance 2015; Business Wire 2012) Energy Efficiency (2016) 9:321–338 323 Computer gaming is engaging an increasingly di- performance requirements of these machines entail verse user base. These consumers spent $22 billion on far higher energy intensities, and in many cases, gaming software in 2013 (ESA 2014), with the global multiple components (e.g., GPUs, hard drives, dis- market estimated at $100 billion (Brightman 2013). The plays) are used. Protocols for benchmarking the scale and growth of this activity calls for assessment of computational performance of gaming computers the associated energy use. involve running a preset gaming process and Just over half of all US households own a game collecting metrics. Some benchmarks focus on cen- console, with the average player being 31 years old tral processor performance (e.g., Cinebench); others and with males and females engaged in roughly equal focus on the graphics (e.g., Unigine Heaven; see proportions. Previous studies exploring the energy im- http://www.maxon.net/products/cinebench/overview. plications of game console use found average unit elec- html and https://unigine.com/products/heaven/). tricity use to be 102 kWh/year for the installed US stock Component product literature, however, emphasizes (excluding the connected display) and 64 kWh/year for nameplate estimates of power requirements, rather new sales as of mid-2012 (Webb et al. 2013).1 There is than actual performance or power needs under a ongoing debate about game console utilization, with given mode of operation. As discussed below, recent studies finding that this may have been previous- accurate energy use calculations cannot be made ly overstated (Desroches et al. 2015). with nameplate data. However, no standardized test We found no prior studies focusing on the aggregate procedures exist for evaluating gaming actual energy used by gaming computers. One assessment computer energy use, which perpetuates market (Ecova 2012) examined the idle power demand of graph- reliance on over-estimates of nameplate data. ic processing units embedded in gaming computers, and The limitations of nameplate data notwithstand- another (Brocklehurst and Wood 2014)exploredwheth- ing, a review of the wide range of nameplate power er these machines would be able to meet the ENERGY requirements for components of analogous perfor- STAR v6.0 requirements, based on pooling diverse test mance already on the market suggests that opportu- results from third-party sources (not standardized for nities exist for improved energy efficiencies in each factors such as choice of motherboard, duration of sleep component, through hardware as well as control im- mode, overclocking, operating system, software running provements (Table 1). A variety of metrics may be during testing, etc.). Their results were confounded by defined for a given component. Useful metrics either differences in test procedures. provide a direct efficiency measure or an analogous This article provides new information based on ratio of energy or power inputs per unit of perfor- nameplate performance of gaming computers and their mance provided. Here, we have picked metrics that components together with direct measurements. Effi- are either industry standards or otherwise readily ciency opportunities are identified.