Content uploaded by Julien Lukasewycz
Author content
All content in this area was uploaded by Julien Lukasewycz on Apr 30, 2025
Content may be subject to copyright.
Achieving Green Computing in Video Games Using Power
Measurements and Organic Computing Techniques
Julien Lukasewycz
University of Duisburg-Essen
Duisburg, Germany
julien.lukasewycz@uni-due.de
Abstract
Video games are currently mostly balanced between the attributes
of performance, holistic quality, and eciency. We propose the
addition of a fourth attribute to be included in this balancing dur-
ing development and usage, namely sustainability. Our approach
addresses this additional attribute by focusing on four main topics:
energy awareness, informedness, self-adaptability, and customiz-
ability. By being able to measure the power consumption of video
games and making them energy-aware, they can adapt to dierent
environments, use cases, and devices. Additionally, the increased
knowledge and visualization of power consumption helps develop-
ers and players to gain a deeper understanding and optimize video
games more eectively towards energy eciency. Customizability
concepts are used to personalize and change the behavior with
regard to the self-adaptation of the video games. All individual
topics combined yield a broadly applicable approach that decreases
power consumption and, hence, increases the sustainability of video
games.
CCS Concepts
•Software and its engineering
→
Software design tradeos;
Power management;•Human-centered computing
→
Infor-
mation visualization; •Applied computing
→
Computer games;•
Social and professional topics →Sustainability.
Keywords
green computing, green gaming, organic computing, power con-
sumption, power management, power visualization, sustainability,
video game engines, video games
1 Introduction
In the last decades, it became the de facto standard to aim for higher
performance and visual quality in video games. Hardware man-
ufacturers tried to enable this development by developing more
powerful hardware components capable of matching the increas-
ingly demanding requirements. However, this aim for increased
performance and visual quality leads at the same time to increased
resource usage and power consumption of devices [
6
,
7
]. Besides
the usage of more powerful hardware components, developers and
manufacturers tried to achieve higher performance through the
more eective usage of resources and optimized algorithms. This ef-
ciency may lead to the misconception that video games are equally
sustainable. However, having no upper limit on resource usage, but
rather using all available resources and only being limited by the
capabilities of the hardware, does not lead to sustainability no mat-
ter how eciently resources are used. This is especially harmful if
the resource usage and power consumption scale faster than the
perceived holistic quality of the video game, leading to a worse
cost-benet factor. This misconception, that sustainability is al-
ready achieved, might also be supported by the lack of detailed data
foundations and statistics, evaluating the sustainability of video
games. Not only high-end video games should be the focus of power
optimizations; even less demanding but unoptimized video games
can lead to a high waste of power when scaling them up to thou-
sands or even millions of players. Based on this, it becomes visible
that even small changes can already have signicant implications
when scaled to the number of players and devices or the time spent
playing a video game.
To introduce the research eld of green computing to this of
video games, a new research eld was established under the com-
bined term “green gaming”. Video games are one of the most de-
manding and interactive types of everyday software on consumer
devices and achieve widespread popularity [
3
]. Based on this, they
constitute a special application scenario in the research eld of
green computing, which is currently not fully covered by other
approaches in this eld and rather overlooked [
5
–
7
]. In our work,
we want to contribute to this young and emerging research eld
by using energy awareness and organic computing techniques to
achieve better and more appropriate resource usage, ultimately
increasing the sustainability of video games. Organic computing
techniques hereby describe methods that let self-organized sys-
tems dynamically adapt themselves to changing environmental
conditions [8].
2 Goals
Video games are usually built to oer the best performance and
holistic quality possible on all platforms. Some platforms, like smart-
phones and handhelds, also demand a high eciency of video
games, as otherwise performance cannot be achieved with limited
resources. Only focusing on one of the attributes would likely al-
ways deem a video game unplayable and decrease the experience of
the player. This, therefore, leads to the development of video games
that are mostly balanced somewhere between the performance,
holistic quality, and eciency attributes. We visualize this balance
with the highlighted triangle in Figure 1, which is inspired by the
iron triangle [
2
]. For now, we will not further dene these attributes
but rely on their common understanding, as it is sucient to follow
our proposed work.
Video games running on devices like smartphones or handhelds
receive the primary focus of eciency research due to the limited
amount of resources available. However, on other devices where
resource scarcity and power consumption are not a major concern,
like PCs or consoles, the primary focus still lies on performance
Lukasewycz
Performance
Efficiency
Sustainability
Holistic Quality
Figure 1: Video Game Attribute Balancing
and less on energy eciency. Our motivation is to increase aware-
ness and knowledge about these underrepresented platforms and
video games in general, as they might be a large source of power
consumption in the software landscape [
6
,
7
]. This is currently only
an assumption, as no broad data foundation and baseline about
the power consumption of video games exists, which we want to
establish with our work. Based on this data foundation, we plan
to propose concrete concepts and practical approaches to achieve
more ecient and appropriate resource utilization, thus reducing
power consumption. This is achieved either by decreasing the re-
source usage without a loss in perceived holistic quality, or by losing
some holistic quality for the benet of even higher power savings.
However, our goal is not to only decrease the frames per second
(FPS) a game is computing. We rather want to dynamically adapt
the game to its content and environment, thus optimizing power
usage and avoiding cases where energy is wasted without any ben-
et. In other words, we argue that eciency and sustainability are
closely related and can only be researched in combination. These
concepts and approaches shall additionally concentrate on both the
developers and players as important stakeholders in this eld. In
total, our goal with this work is to measure and optimize the power
consumption of video games to increase sustainability. To achieve
this, we want to introduce sustainability as an additional attribute
to be balanced with performance, holistic quality, and eciency. To
visualize this, we extend Figure 1 by the newly introduced attribute.
To clarify the focus of our work, we propose the following re-
search questions (RQs):
RQ1
How can the power consumption of video games be mea-
sured?
RQ2
How high is the power consumption and eciency of in-
dividual components (e.g., graphics or physics) inside video
games?
RQ3
Which technical factors inuence the power consumption of
video games?
RQ4
How can we inform and educate stakeholders about the
power consumption of video games in the most benecial
way?
RQ5
How can video games and video game engines integrate
organic computing techniques to adapt to dierent environ-
ments and usage scenarios, resulting in a more sustainable
and reduced resource usage?
RQ6
How can stakeholders be enabled to customize the power
consumption behavior of video games?
3 Approach
To answer our previously dened research questions, we rst want
to generate measurement data about power consumption to create
a baseline of current energy eciency in video games. Based on this
baseline, we can propose optimizations and customization options
using organic computing techniques that allow for more ecient
resource usage. Additionally, we can use this data to inform and
educate stakeholders, in our case developers and players, about
power consumption. We therefore divide our work into the follow-
ing topics: energy awareness, informedness, self-adaptability, and
customizability. In the following, we describe each topic in detail.
3.1 Energy Awareness
Before we can optimize the power usage and eciency of video
games, we rst need to establish a baseline of current power con-
sumption. There are a variety of dierent hardware and software
methods and tools already in place that allow the measurement
of power for a whole device, a single application, or on a function
level [
4
]. However, none of them oers a comprehensive or adapted
solution for the interactive nature and application scenario of video
games. This results in a large eort and impracticalities for the
stakeholders to eectively utilize these tools. This is caused by
measurement results being imprecise, as they are not semantically
related to the corresponding part of a video game. Based on this, we
rst want to propose a universal method and provide tool support
to measure the power consumption of video games adapted to the
use cases of developers and players. This will answer RQ1. Our
method shall measure a video game on all levels, ranging from the
overall consumption down to that of a single functionality while
providing meaningful context to better optimize video games for
power consumption. This enables developers to better measure and
optimize their game for energy eciency, as well as provide useful
information to players. To broaden the picture of power usage, this
measurement might also be extended to additional but underrep-
resented hardware like, for example, monitors that count into the
overall consumption of PC gaming.
After dening our method and creating the software to measure
power consumption in video games, we want to use it to build
a data foundation. We plan to measure slices of concrete video
games to analyze where power is consumed and which factors
inuence the amount of power needed, thus answering RQ2 and
RQ3. Slices hereby refer to representative and reproducible aspects
of a video game like gameplay sequences, cutscenes, or menus.
We aim to measure video games across dierent types, genres,
platforms, engines, and settings to achieve broad and general results.
This will enable us to create a broad data foundation and actual
statistics about the total power consumption of individual video
games as well as gaming in general, which could until now merely
be approximated [6].
3.2 Informedness
When we are able to measure the power consumption of video
games, we also want to provide this information in a useful form
to the stakeholders. This information might dier for players and
developers, as developers might need far more detailed information
Achieving Green Computing in Video Games Using Power Measurements and Organic Computing Techniques
to optimize their video games, while players often have fewer possi-
bilities to adapt their video games. Here, we want to research which
information is needed for which party and which approach is most
benecial to visualize and convey the information about power
consumption. This answers RQ4 and ultimately also increases con-
sumer information [3]. Additionally, we want to implicitly trigger
a learning eect by the stakeholders. This eect could be achieved
by presenting eects on power consumption directly to the stake-
holder; for example, the resulting power consumption if settings for
a higher or lower visual quality are selected. Using the previously
described measurement data in combination with estimations or
benchmarks, we can determine how power consumption would
change. In this way, we can visualize an association between de-
cisions and their consequences in power consumption, possibly
resulting in additional considerations to optimize power consump-
tion for a suitable cost-benet factor. By educating stakeholders
about power consumption in real-time and personalizing it to their
use case, we believe that we can achieve higher awareness and thus
a possible reconsideration of priorities [3].
3.3 Self-Adaptability
With the knowledge of power consumption from Section 3.1, we
not only want to make video games energy-aware but also self-
adapting, based on the state of themselves and their environment.
Currently, most functionalities inside a video game are not gener-
ally limited in their resource usage or execution times, except the
upper limits posed by the underlying device or video game engine.
This results in the competition of dierent functionalities for more
and more resources and execution time. It also means that some
functionalities might be part of this competition, which yield no
benet to the overall performance or holistic quality of the video
game when receiving more resources. For example, it might not
be sensible to allow the physics engine of a video game to use the
same amount of computational resources for explosion particles in
an action scene and in the main menu. The latter would obviously
bind resources that might not positively inuence the perceived
quality of the game and could be assigned to other functionalities
or partially freed. In well-engineered video games, this would not
happen, as developers manually adapt and limit the resource usage
of their functionalities. To ease the eort of developers, we want
to use organic computing techniques that automatically delegate
resources to where they are needed based on the task at hand. This
means that the game and the currently computed in-game scenes
are analyzed to understand which functionalities are processed.
As video games are a very dynamic medium, these functionalities
and their needed resources can vastly dier within a few rendered
frames. This makes it crucial to constantly monitor the current
state of the game and assign resources accordingly to where they
are needed. It might be possible to use the concept of task sched-
ulers for this problem; however, based on the dynamic nature of
video games, they might not be suitable for this type of software
as upcoming tasks would be hardly predictable. An adapted form
of task scheduler or another approach might therefore be needed
to be used in video games. Additionally, we only want to assign
the amount of resources needed to fulll the given task suciently.
This contradicts the current aim to render or compute as many
cycles of execution as possible per second, even if they are objec-
tively not needed to achieve a good player experience. We do not
only want to adapt to the content of the video game but also to the
environment, in particular to the device the video game is executed
on. Today, most video games are developed as a single code base,
which is then ported and optimized for dierent platforms such as
PCs, consoles, or mobile devices. Besides some consoles and hand-
helds, most video games are, however, not developed for a specic
hardware conguration. Especially in the PC market, it would also
be impossible to optimize a video game for each individual com-
bination of hardware components. We want to make video games
adaptive to the specic hardware components of the device. This
could, for example, mean that video games automatically adapt to
the read and write speeds of storage devices. In the case of a slower
HDD, the video game could automatically request les earlier, thus
adapting to this hardware component. Both, the self-adaptation
based on the game itself and the environment, would optimize the
resource usage and, thus, the performance and power consumption
of video games, ultimately answering RQ5.
3.4 Customizability
Finally, we want to empower the stakeholders to directly inuence
the self-adaptation and power usage behavior, described in Sec-
tion 3.3, by providing their intentions and preferences. Developers
could, for example, congure the initial power usage behavior of
the video game during development. Players could then further
customize these initial congurations to better match their device,
environment, and preferences. This could range from extending
the concept of favoring visual quality or performance, often known
from consoles, to detailed power settings directly integrated into
the game. This is especially important as Mills et al. showed that
player decisions inuence power consumption more than technol-
ogy choices [
6
]. With this last concept, we can take into account
factors from the video game itself, the underlying device, and the
stakeholders to achieve better sustainability in video games. This
approach helps us to answer RQ6.
4 Challenges
Our approach proposes the eective and reduced usage of resources,
leading to more sustainable video games in the process. It is evi-
dent that to achieve this goal, compromises need to be made and
restrictions need to be implemented. As currently the focus in
the industry lies on achieving the highest performance in video
games, our proposals will most likely also include some kind of
performance reduction. Mills et al. identied that the industry, and
players alike, mostly have little fundamental interest in the topic
of sustainability of video games, leading to a possible lack of re-
search cooperations [
6
]. It might therefore be necessary to switch
focus towards developing concepts that are easily applicable by
developers and oering near identical holistic quality to players
while being more energy ecient. This might lower the barriers for
the stakeholders to get involved in the topic. Moreover, additional
benets need to be identied that emphasize the development of
more sustainable games by choice and not only by the possibility
of money savings or complying with policies and laws. While the
Lukasewycz
interest in sustainable games generally increased slightly over the
last few years, this challenge might still need consideration.
Another challenge lies in the access to the source code of video
games and their engines. Nearly all video games and most engines
use a closed-source approach to protect their business model against
competitors and piracy attempts. In our case, it restricts us from
generating measurement data, as well as integrating and evaluating
our proposed concepts in both video games and video game en-
gines, as the source code would be necessary. We, therefore, strive
for collaboration with video game and engine developers to pro-
vide benets and shared knowledge to both industry and research.
However, concepts that do not rely on the availability of the source
code are also part of our research.
5 Current Progress & Completion Plans
Currently, we are actively working on the power measurement
method described in Section 3.1 as our next publication. While the
theoretical method is already dened, the prototype to highlight
the applicability of our method is still under development. Our
plan is to publish this work after nishing the last remaining work
items in the next few months. Based on this initial foundation,
we are then able to focus our eort towards the other described
topics. Especially, the measurement of concrete video games and
the presentation of power usage data to developers and players as
described in Sections 3.1 and 3.2 represent the planned focus of our
work during the next year.
6 Related Work
We described a variety of dierent topics and approaches. To put
these into perspective, we want to highlight a selection of other
research works targeting similar problems.
To measure the power consumption of software in general, dier-
ent approaches and tools can be used. Ghaleb summarized dierent
hardware and software tools and their individual capabilities in
their work [
4
]. It is apparent that there is no all-in-one measurement
solution that would cover all hardware components, application
scenarios, and measurement levels. This makes it too complicated
and eort-intensive to eectively utilize these tools in video game
development. Players using video games will also unlikely invest
time and money in additional measurement hardware or software.
We therefore argue that it is particularly important to oer a tailor-
made solution for video game developers and players that works
out of the box without the need for additional hardware or cong-
uration.
In their project report, Mills et al., among other things, high-
lighted the importance of ecient gaming concepts as gaming
causes a high percentage of residential power consumption [
6
].
They showed that PC gaming alone consumed 4.1 terawatt-hours
of electricity and generated 1.5 million tons of CO2-equivalent emis-
sions in 2016 just in the state of California. These values give an
initial impression about the impact factor of video games; however,
their global impact is still unclear and can merely be approximated
based on these results. Similar to the authors, we also believe that
rst a baseline of data must be established before optimizations
can be implemented, as otherwise no baseline, to compare against,
exists. They argued that more mature measurement techniques are
needed to generate this missing data. The authors also provide a
detailed look at dierent measurements they conducted, revealing
important results and learnings for developing more sustainable
video games.
In 2009, Anand et al. proposed a self-adaptation principle that
relies on player activity [
1
]. This means the video game uses more
resources if a high player interaction is noticed, and power-saving
measures if the player interacts less with the video game. Based
on a linear prediction algorithm using error weights as a feedback
loop, they achieved mostly identical player experiences while re-
ducing the power consumption by turning o the network card.
The authors suspected a connection between the amount of player
interaction and network communication transfers, which oered a
starting point for power optimizations. Our approach focuses less
on the player but more on the game itself, as we try to identify
a dierent connection, namely between the displayed content of
the video game and possible power savings. However, adapting the
video game to the presence and interaction amount of the player
might also lead to more self-adaptation behaviors in our concept.
In the preparation of this work, we have, to the best of our
knowledge, not found any research that specically targets the
power usage customization of video games. This could mean that
this topic was not yet applied to the eld of video games, but
relevant knowledge could likely still exist in the general eld of
green computing.
7 Conclusion
In this paper, we have described an approach to measure and op-
timize the power consumption of video games to increase their
sustainability. Reduced power usage will not only reduce the envi-
ronmental impact but also the cost of gaming, especially in areas
with high energy prices. Our approach consists of a concrete mea-
surement method that will be dened and used to identify possible
high consumption factors in video games. Players and developers
shall be educated about these factors so they can in turn use this
knowledge in development and when gaming. Not only the stake-
holders shall adapt their video games, but also the video games
themselves shall adapt using organic computing techniques. With
these concepts, video games can better allocate resources and adapt
to their environment and device, where they are executed on. This
self-adaptability is then extended with the intents of the stakehold-
ers, making it customizable to personal preferences. We believe that
this combined approach will be an essential rst step to reduce the
environmental impact of video games. Especially as the increased
knowledge about power consumption could lead to changed behav-
ior and processes in the development and usage of video games.
References
[1]
Bhojan Anand, A. L. Ananda, Mun Choon Chan, Long Thanh Le, and Ra-
jesh Krishna Balan. 2009. Game action based power management for mul-
tiplayer online game. In Proceedings of the 1st ACM Workshop on Network-
ing, Systems, and Applications for Mobile Handhelds (Barcelona, Spain) (Mobi-
Held ’09). Association for Computing Machinery, New York, NY, USA, 55–60.
https://doi.org/10.1145/1592606.1592619
[2]
Roger Atkinson. 1999. Project management: cost, time and quality, two best
guesses and a phenomenon, its time to accept other success criteria. International
Journal of Project Management 17, 6 (1999), 337–342. https://doi.org/10.1016/
S0263-7863(98)00069- 6
Achieving Green Computing in Video Games Using Power Measurements and Organic Computing Techniques
[3]
Ana Cardoso. 2020. Reducing the energy use of video gaming: energy eciency and
gamication. Technical Report. Copenhagen Centre on Energy Eciency.
[4]
Taher Ahmed Ghaleb. 2019. Software Energy Measurement at Dierent Levels of
Granularity. In 2019 International Conference on Computer and Information Sciences
(ICCIS). 1–6. https://doi.org/10.1109/ICCISci.2019.8716456
[5]
Evan Mills, Norman Bourassa, Leo Rainer, Jimmy Mai, Arman Shehabi, and
Nathaniel Mills. 2019. Toward Greener Gaming: Estimating National Energy
Use and Energy Eciency Potential. The Computer Games Journal 8 (12 2019).
https://doi.org/10.1007/s40869-019- 00084-2
[6]
Evan Mills, Norman Bourassa, Leo Rainer, Jimmy Mai, Ian Vaino, Claire Curtin,
Louis-Benoit Desroches, and Nathaniel Mills. 2019. A Plug-Loads Game Changer:
Computer Gaming Energy Eciency without Performance Compromise. Technical
Report. Lawrence Berkeley National Laboratory.
[7]
Nathaniel Mills and Evan Mills. 2016. Taming the energy use of gaming computers.
Energy Eciency 9, 2 (01 Apr 2016), 321–338. https://doi.org/10.1007/s12053-015-
9371-1
[8]
Christian Müller-Schloer, Christoph von der Malsburg, and Rolf P. Würt. 2004.
Organic Computing. Informatik-Spektrum 27, 4 (01 Aug 2004), 332–336. https:
//doi.org/10.1007/s00287-004- 0409-6
