However, all these games, albeit virtual, are run on physical hardware, and therefore consume energy. This article presents and compares the energy consumption of different video games and their parameters. To find out how much energy these uses actually consume, we have chosen to evaluate the following video games: Assassin’s Creed Valhalla, Total War Warhammer III, Borderlands 3, Anno 1800 and War Thunder.

We have previously carried out a study on mobile games.

Selection and methodology

These video games were selected because they offer a benchmark. Using these benchmarks as measurement subjects ensures the replicability of our experimental protocol, while eliminating the human factor from the results.

A benchmark is a feature offered by the game that allows you to measure the performance of a system (entire PC), or one of its components (CPU, GPU, memory, etc.) according to a given scenario and selected parameters.

We’ve also taken care to represent several types of game mode, such as RPG (role-playing game), strategy or simulation.

We measured these video games on a PC with the following configuration:

• Processor: i7 6700 
• Memory: 32 Go RAM DDR4 
• Graphics card: RTX 3060 12Go 

This equipment was supplied to us by OPP!, a company offering PC and Mac repair and maintenance services, as well as individual component sales.

The screen used is an LG E2441 with the following specifications:

• Screen technology : LED 
• Screen Size : 24” 
• Resolution : 1920×1080 

We collected energy metrics using a measurement module connected to our Greenspector Studio software, plugged directly into the PC and monitor power supplies and connected to the mains socket.

Benchmarks were carried out in 2 different graphics configurations:

• A configuration with maximum settings for the graphics offered by the game
• A configuration with minimal settings for the graphics offered by the game

6 iterations were performed on each scenario to ensure reliable results.

Benchmarks last between 80 and 240 seconds. These variations do not affect the results presented.

Graphic evolution impacts power

Modern games incorporate higher-quality graphics with ultra-detailed textures, advanced visual effects such as dynamic lighting, real-time shadows and sophisticated particle effects. This graphical complexity requires considerable rendering and graphics processing capabilities.

Gamers are also increasingly opting for high display resolutions for an optimal visual experience. This puts extra pressure on the GPU (Graphics Processing Unit) to render detailed images at ultra-high resolutions.

These GPUs have increasing energy consumption with each new generation, as shown below for NVDIA:

Evolution of minimum system power and maximum GPU power by GPU release date

Developers exploit advanced rendering techniques such as ray tracing to realistically simulate the behavior of light in virtual environments. Although these techniques offer an unprecedented level of realism, they are computationally intensive and require high-end GPUs.

Consumption differences depending on settings

Measurements of average PC power on the lowest and highest graphics configurations for each game show a wide disparity between them.

Total PC power at minimum or maximum setting

Switching from the maximum settings to the lowest settings offered by each game results in a measured power reduction of 45% on average. In the case of Borderlands 3, a power gain of 72% can even be observed.

In Anno 1800, the benchmark is a panoramic aerial view of the game’s map. This sequence highlights details of the game world, such as landscapes, iconic buildings and animations of everyday life.

Below are graphs of one iteration measured with maximum parameterization and another iteration with minimum parameterization. The benchmark sweeps over the city from its zoomed-in aerial viewpoint at the start, then the same trajectory is repeated 8 times with increasingly higher viewpoints, which explains the 8 peaks on the graph.

Here, we can easily see the significant difference between the 2 parameter levels. On the two different settings, we can see first of all that the further the camera is from the city, the more the power is reduced, given the increasingly short time of the scenario.

What’s more, when the game is set to maximum, power consumption is at its peak for almost the entire duration of the scenario, whereas measurements taken with the lowest setting show lower and shorter power peaks.

Anno 1800 benchmark power consumption with maximum settings

Anno 1800 benchmark power consumption with minimum settings

A Statista survey conducted in December 2023 revealed that 22% of US adults aged 18 to 29 spent six to ten hours a week playing video games. Overall, respondents in this age group were also more likely than others to be avid gamers, as a total of 8% played video games for more than 20 hours a week on average.

These figures enable us to evaluate overall energy consumption over the usage times of different types of players, in the case where the benchmark is representative of game consumption. Energy consumption has been projected with measurements taken on the minimum and maximum settings of each game.

The average consumption for one hour of play at minimum settings is 0.168 kWh, and 0.254 kWh at maximum settings. These results are higher than those of the European study on the environmental impact of digital services. The latter shows a consumption of 0.137 kWh for one hour of PC gaming at medium resolution.

Most gamers therefore consume between 1.5 kWh and 2.5 kWh per week, playing between 6 and 10 hours a week. For more involved gamers playing around 20h a week (2h40 a day), their PC and screen have a weekly consumption of 5 kWh. By the same token, a conventional refrigerator consumes an average of 3.29 kWh per week.

Evolution by release date

On maximum configurations, we note an evolution in measured power proportional to the release date of these games.

In this context, the maximum configurations of video games reflect this technological evolution. Game developers design their games to take advantage of the latest hardware advances, and this translates into increasingly high demands on components. As a result, to take full advantage of graphics performance and game fluidity, gamers often have to invest in state-of-the-art hardware.

These complex, detailed graphics require real-time rendering, which often relies on the CPU to perform calculations related to physics, artificial intelligence of non-player characters, collision management and other aspects of gameplay.

This is what a technical director of the Total War game explains in an interview with Intel:

“We model thousands of soldiers with a high level of detail applied to each in terms of animations, interactions, pathfinding decisions, etc.”

In video games, pathfinding consists in figuring out how to move a character from point A to point B, taking into account the environment: obstacles, other characters, length of paths, etc.

What’s more, the processor is often juggling numerous tasks simultaneously, depending on what’s displayed on the screen. “Take a scene where two huge fronts with thousands of soldiers are smashing into each other, and you’ve zoomed in quite close,” explains the game’s technical director. “In this situation, the processor is divided mainly between entity agent-based combat, collision mechanisms and building matrix stacks to draw all the entities.”

In other words, the processor must simultaneously manage the presence and interactions of thousands of NPCs (non-player characters).

What’s more, the more advanced the graphics, the greater the demand on the GPU to process data and instructions efficiently, which can lead to bottlenecks and slowdowns if the processor isn’t powerful enough.

In the Assassin’s Creed Valhalla game, when the settings are at their lowest, the graphics card is called upon for 46% on average. Conversely, at maximum setting, with water reflections activated or cloud quality at maximum, the graphics card is used at 99% during the benchmark.

Optimization vs. graphic quality

We’ve just seen that setting a game to its maximum setting involves high energy consumption. But are the visual effects enhanced? Are all settings relevant to the gaming experience, depending on the PC configuration?

An interesting indicator to answer these questions is the number of frames per second (FPS), as it is often used as an indicator of a game’s fluidity: the higher the FPS, the more fluid and responsive the game appears.

FPS (Frame Per Second) indicates the number of individual images (or “frames”) displayed on the screen each second.

The more complex an image is to generate and display, the slower the processor and graphics card can display it. So, when the settings exceed the capabilities of the PC configuration, the visual effect for the player is not necessarily enhanced.

What’s more, gameplay can be impacted by the bottleneck phenomenon.

A bottleneck is a phenomenon produced by a hardware or software component with limited performance compared to other, more powerful components. This means that one part of the system is operating at maximum capacity, while other parts can’t keep up, resulting in a drop in overall performance.

By correctly balancing the hardware configuration and adjusting graphics settings accordingly, gamers can minimize the risk of slowdowns and jerks, delivering a more enjoyable and immersive gaming experience.

Here are some differences from benchmarks set to their maximum gaming parameters and then to their minimum:

Implications for materials and environmental impact

The constant evolution of video games towards ever more immersive and realistic experiences has significant implications for the hardware used. Game developers are seeking to fully exploit the graphics and processing capabilities of new technologies, which translates into higher hardware requirements.

In France, 2020 will see the sale of 2.3 million consoles, 27.5 million complete games (console + PC / physical + dematerialized) and almost 7 million accessories (console + PC). With growth of 10%, the console ecosystem represents 51% of the total video game market, while PC gaming is up 9%. (Source: Sell)

Gamers are looking to stay at the cutting edge of technology to take full advantage of the latest releases. Beyond the financial stakes this may represent, this quest for hardware performance is also very critical from an environmental point of view.

As we’ve seen, all the components in a configuration need to be at roughly the same level of performance for an optimized gaming experience. If the gamer has a high-performance graphics card but a lower-resolution screen, or a less powerful processor or motherboard, the gaming experience will not necessarily be enhanced or even altered. So, from an optimization point of view, it’s not so much a matter of buying the latest components to improve performance as it is of optimizing game parameters according to your hardware configuration. On the one hand, this extends their life expectancy through reduced stress, but it also improves the gaming experience for users.

Over-solicitation causes components such as the graphics card or processor to heat up to high temperatures due to the amount of calculations they handle, and damages their transistors and chips, thus shortening their lifespan.

According to HP, the lifespan of an average desktop PC is between 2 and 3 years, and that of a gamer PC between 3 and 5 years.

We have no information about the environmental impact of gamer PC manufacturing, but the frequency of release of the latest generation of products, which pushes gamers to renew their PC configuration every year, considerably increases the impact of this industry.

It’s worth noting that game consoles also have a significant carbon footprint.

Climate and sustainability researcher Ben Abraham analyzes the PlayStation 4’s central processing unit using mass spectrometry, revealing the presence of atomic components such as titanium, whose extraction, refining and manufacture contribute to greenhouse gas emissions.

This observation underlines the challenge of making the production of these devices sustainable, with decades needed to achieve this goal.

The importance of measurement

Video game publishers play a crucial role in reducing the industry’s environmental footprint. To do so, it is imperative to take energy consumption into account throughout the game development process.

First and foremost, measuring energy consumption enables game publishers to understand the environmental impact of their products. This includes not only the direct energy consumption of the devices on which the games are run, but also the carbon footprint associated with game servers, updates and downloads.

Secondly, this awareness enables game developers to design game mechanics and graphics that optimize energy efficiency. For example, by minimizing complex visual effects that require high computing power, games can reduce their energy consumption while still offering an immersive gaming experience.

The subject of the environmental footprint of video games is increasingly being taken on board by publishers, which is encouraging. Initiatives such as Ukie’s Green Games Guide and Ecran d’après offer practical advice and best practices for reducing the environmental impact of game design and development. Similarly, tools such as Microsoft’s Xbox Sustainability Toolkit or Jyros, the environmental impact measurement tool dedicated to the video game industry in France, provide developers with concrete ways of assessing and improving the sustainability of their games.

However, it is important to generalize these practices and integrate them more systematically throughout the industry. Too often, the environmental aspect is relegated to the background, while the emphasis is placed on the performance and aesthetics of games. It is therefore essential that publishers take greater account of the environmental implications of their design and development decisions.

Study limits 

In the context of this study, it is important to recognize certain limitations that could affect the scope and representativeness of the results obtained:

1. Manufacturer/designer partnerships: It is possible that some video games have established partnerships with hardware manufacturers to optimize their performance on specific configurations. These agreements could distort benchmark results by favoring certain brands or models of components. These results may alter comparisons between games, but not comparisons between configurations within the same game.

2. Benchmark scenario not necessarily representative of game modes: Benchmark scenarios used to evaluate video game performance may not reflect actual gameplay conditions. For example, a benchmark may focus on specific game sequences that do not necessarily represent general gameplay. As a result, the results obtained may not be fully representative of the overall gaming experience.

3. No measurement of multiplayer or online play: This study focuses primarily on the performance of single-player games, and does not take into account aspects related to multiplayer or online play. Consequently, data exchanges between game servers and clients, as well as network performance, are not taken into account in the analysis. This could limit a complete understanding of hardware requirements for an optimal online gaming experience.

Conclusion  

In conclusion, this study highlights the growing impact of video games on computer hardware performance. With the constant evolution of graphics and functionality, modern games require increasingly powerful hardware configurations to deliver an optimal gaming experience. This raises important questions about the sustainability and energy efficiency of computer equipment, as well as consumer hardware choices. Ultimately, it is crucial for both publishers and gamers to strike a balance between video game performance and the sustainability of the technology industry to ensure a more sustainable future.

What do printers, connected cars and airliners have in common?

In this article, we will mention a few stories of surprising computer attacks. This will enable us to question our choices when it comes to implementing new features. These misadventures all have a common cause: an increase in the attack surface.

The multiplication of access points is a risk factor

In recent years, we’ve all seen objects that communicate with the outside world appear in our living rooms. From connected voice assistants to smart thermostats, these objects provide more or less useful services. The business world is no exception to this rule. Whether as part of the Industry 4.0 vision, or simply to facilitate remote communication, these connected systems are playing an increasingly important role.

Unfortunately, some devices pose major risks. Combining a low level of security with a connection to a company’s internal network, connected objects are a goldmine for malicious individuals. And they don’t hold back.

In its article, 01net shows how a group of Russian hackers is attacking connected objects to target businesses: https://www.01net.com/actualites/un-groupe-de-hackers-russes-cible-les-objets-connectes-pour-s-attaquer-aux-entreprises-1743886.html

What’s more, these connected objects often have access to private data. Imagine someone turning on your webcam remotely, or accessing the microphone on your soup mixer. Worse still, imagine a malicious individual taking control of a child’s toy and using it to contact him or her: https://www.france24.com/fr/20170228-hackers-ont-pirate-peluches-connectees-fait-fuiter-messages-denfants-a-leurs-parents

The proliferation of these objects poses a real social problem that we cannot ignore.

From an environmental point of view, the distribution of these systems also has significant impacts. From mineral extraction to distribution, the production of IT systems generates significant CO2 emissions, not to mention other impacts such as soil pollution and the erosion of biodiversity.

For all these reasons, the purchase of a new connected device should not be taken lightly. The question is: do we really need it?

How can an ancillary feature turn into a Trojan horse?

New connected objects aren’t the only systems that can be attacked: existing software can be as well.

Nor is it just a question of resources. Aviation, one of the world’s most financially powerful industries, which has invested considerable resources in security, has also been the victim of criminal acts.

In this article, we won’t be discussing the impacts of flying, but rather the specific subject of in-flight entertainment.

The many films and series available bring undeniable benefits for users: boredom reduction, keeping children occupied, forgetting about stress (and the fact that you’re in an aircraft that burns thousands of liters of fuel per hour) …

Nevertheless, the screen is not a system totally isolated from the rest of the world. For example, cutting video during a staff call necessarily implies communication between the box and at least part of the rest of the device.

And this link can be used to support an attack.

Chris Roberts, a cybersecurity specialist, has demonstrated this by successfully modifying the power of a reactor using the entertainment system: https://www.01net.com/actualites/un-hacker-aurait-pris-le-controle-d-un-avion-en-vol-grace-a-son-systeme-de-divertissement-654810.html

In reality, it’s extremely difficult to totally isolate one system from another.

This story is just one example:

• Connected cars: https://securite.developpez.com/actu/334250/Une-faille-dans-les-voitures-Tesla-permet-aux-voleurs-d-avoir-leur-propre-cle-et-de-s-emparer-de-la-voiture-en-seulement-130-secondes-l-exploit-tire-parti-du-lecteur-de-carte-NFC-des-voitures-Tesla/

• Facebook ancillary functionality : https://tech.hindustantimes.com/tech/news/facebook-data-breach-what-is-view-as-feature-and-why-it-s-been-disabled-story-vpL6KmzapfYxZVwQBumNyI.html

This last attack is an interesting one. It illustrates the issue of a well-known developer philosophy: “Why do it? Because we can.”

Hackers have taken advantage of a security flaw in a service of Meta’s flagship social network. The functionality in question allowed users to see how their profile was viewed by another user. Admittedly, this is of interest to the user, but it is not essential to the smooth operation of the social network. On the other hand, the consequences of an attack are extremely damaging, both for users and for the company, whose image is tarnished.

When the group became aware of the flaw, they immediately removed the service. The question then arises: did users notice the disappearance of the functionality?

From a general point of view, we can list a few disadvantages of the multiplication of possibilities offered by a digital service:

• dispersion of resources that could have been allocated to securing key application or website services
• implementation of little-used functionalities that receive little attention from the development team and are therefore more vulnerable
• the need to reduce compatibility with older versions of Android or iOS. And consequently reduce the number of potential users
• increase the weight of an application due to the development of more code or embedded media. Increasing the application’s environmental impact.

Taking into account the associated risks, we must always ask ourselves: is the comfort it brings really worth the impact it causes?

It’s also worth remembering that cybersecurity is an integral part of digital sustainability. As a designer of digital services, it is therefore our duty to protect users. Implementing security mechanisms is an important part of this, but we also need to think globally, encompassing all functionalities.

Malicious individuals will try to get into every nook and cranny of your system. By increasing the number of functions, you are giving them new doors that they will be happy to open.

Finally, all these attacks show us that digital sufficiency is not only a useful tool in the context of the ecological transition, but is also of interest in the fight against cybercrime.

Conclusion 

In short, digital sufficiency is proving to be our unexpected ally in the daily battle for IT security. Before rushing off to buy the latest gadget or design a new feature, let’s ask ourselves the following 2 questions:

1. Is that useful ?

2. Is the risk worth the benefit?

In some cases, the answer is obviously yes. The seatbelt makes the car heavier and therefore increases fuel consumption, but it considerably reduces the number of deaths on the roads. The reduction in comfort was worth it.

In many cases, the answer is the opposite. Today’s cars can reach speeds well in excess of 150km/h. Yet it is forbidden to exceed 130km/h. This measure, taken in France in 1974 to combat the 1973 oil crisis, was the result of a balancing act between individual freedoms on the one hand, and the collective effort needed to counter the consequences of the oil crisis on the other. It wasn’t worth the risk.

This central consideration in any decision must be at the heart of a development team’s questioning.

Today, only the advantage part of a feature is highlighted. But that’s forgotten:

• User security
• The financial cost of a computer attack
• Damage to the image of the company that suffers a computer attack
• The environmental impact of this functionality
• Loss of compatibility with certain users
• And many more…

33 years after the introduction of compulsory rear seat belts, the question of discomfort versus safety is no longer an issue in the automotive world. It must also become a reflex for digital service design teams in the IT world.

In France, the accessibility of digital services has had a legislative framework for several years now (initiated by article 47 of the 2005-102 law of 11 February 2005 [FR] and specified in decree no. 2019-768 of 24 July 2019 [FR]). This is based primarily on the RGAA [FR] (Référentiel Général d’Amélioration de l’Accessibilité – General Accessibility Improvement Reference Framework). The eco-design of digital services, which has been discussed in France for over 15 years, has gained considerable momentum in recent years. However, the subject is still struggling to establish itself, or even to take precise shape within organisations. The legislative framework has been taking shape since 2021 and should enable the eco-design of digital services to take hold over the next few years. The aim of this article is to shed some light on the subject.

A quick reminder

ADEME (Agence de l’Environnement et de la Maîtrise de l’Energie) and ARCEP (Autorité de régulation des communications électroniques, des postes et de la distribution de la presse) are working together on the environmental impact of digital technology. Their work covers, in particular, the estimation of these impacts on a French scale, as well as best practices and prospects. This information can be found here: https://www.arcep.fr/nos-sujets/numerique-et-environnement.html [FR]

Ecodesign [FR] can be defined as an approach that integrates the reduction of environmental impacts right from the design stage of a digital service, with a global vision of the entire life cycle, via continuous improvement.

A digital service [FR] is a set of human, software and hardware resources needed to provide a service.

Consequently (but we’ll come back to this in a later article), talking about an eco-designed website can be perceived as a misuse of language. As part of an eco-design approach, we need to take an interest in all the site’s digital services (or at least a representative sample), through continuous improvement and by covering all the stages in the project’s lifecycle. All this goes much further than simply measuring a sample of pages on a site that is already online.

The laws

In France, there are currently 2 main laws: the AGEC law (Anti-Gaspillage pour une Économie Circulaire) and the REEN law (Réduction de l’Empreinte Environnementale du Numérique).

The AGEC law [FR] briefly addresses the subject, but this requirement does not yet seem to have been dealt with exhaustively. On this subject, see the Guide pratique pour des achats numériques écoresponsables from the Mission interministérielle Numérique Écoresponsable [FR].

Even if certain elements still need to be clarified, the REEN law [FR] goes further by mentioning (among other things) :

• The need to train engineering students in digital-related courses in the eco-design of digital services. But there is also a need to raise awareness of digital sobriety from an early age.
• The creation of an observatory on the environmental impact of digital technology, via ADEME (Agence de l’Environnement et de la Maîtrise de l’Énergie) and ARCEP (Autorité de régulation des communications électroniques, des postes et de la distribution de la presse).
• A general reference framework for the eco-design of digital services to set criteria for the sustainable design of websites, to be implemented from 2024. ARCEP has since confirmed that this benchmark will be based on the RGESN (Référentiel général d’écoconception de services numériques [FR]): https://www.arcep.fr/actualites/actualites-et-communiques/detail/n/environnement-091023.html [FR] A public consultation, launched in October 2023, aims to consolidate this benchmark and practices around it, with a view to wider adoption from early 2024.
• The fight against the various forms of obsolescence, as well as actions to promote re-use and recycling.
• Reduce the impact of data centres (in particular by monitoring the efficiency of energy and water consumption) and networks. The decree is currently being published [FR].
• Require municipalities and groups of municipalities with more than 50,000 inhabitants to draw up and implement a Responsible Digital Strategy by 2025. This strategy must include elements relating to the eco-design of digital services. A number of guides have been published to help establish this strategy, including this one: https://www.interconnectes.com/wp-content/uploads/2023/06/web-Guide-methodologique_V8.pdf [FR]

All of this is accompanied by the establishment of the HCNE (High Committee for Eco-responsible Digitisation), various roadmaps and an eco-responsible digital acceleration strategy. All this is detailed on this page: https://www.ecologie.gouv.fr/numerique-responsable [FR]

What’s next?

Once all these elements have been defined, the question arises of what remains to be done.

In 2024, the REEN law will require public websites to be designed in a sustainable way. By 2025, local authorities with more than 50,000 inhabitants will have to have integrated this dimension into their Responsible Digital Strategy.

Greenspector has been involved in the eco-design of digital services for several years. This evolution in the legislative framework coincides with our involvement in projects at an increasingly early stage, sometimes even from the expression of need. This inevitably requires changes in practices, including the introduction of ideation workshops that take into account the environmental footprint of a service. More and more often, the RGESN is used as a reference to guide the approach throughout the project. This reference framework is ideal for this type of support, but it also provides a basis for managing eco-design as a continuous improvement process.

This way of rethinking support for the eco-design of digital services also makes it possible to move towards greater impact reduction levers and to involve more types of profiles in the projects supported.

As the process begins with public institutions, it is to be hoped that companies will follow suit. In fact, some have already begun the process of complying with the RGESN. Not just in anticipation of a possible change in the legislative framework affecting them, but also because these standards provide a long-awaited framework for the eco-design approach.

To support all these efforts, financial aid is available for both companies [FR] and local authorities [FR].

On all the issues raised here, France has made great strides. Now it’s up to other countries to follow suit. In September, the W3C (World Wide Web Consortium) published its WSG [FR] (Web Sustainability Guidelines). They are now out for public consultation with a view to making further progress on the subject and perhaps eventually establishing web standards. They are also accompanied by discussions on the best way to introduce levers directly at institutional level. In Europe, some countries, notably Belgium and Switzerland, are federating around structures similar to the INR. It is to be hoped that the RGESN and other elements currently in place in France can be adapted to other countries.