RT journal article
T1 Green Gaming: Automated Energy Consumption Reduction for Doom Engine
A1 Aragón Jurado, José Miguel
A1 Ruiz Villalobos, Patricia
A1 Dorronsoro Díaz, Bernabé
A1 Thawonmas, Ruck
A2 Ingeniería Informática
AB The video games industry dominates the entertainment market, standing among one of the most significant sectors in the world. Its continuous and rapid growth, driven by both the expanding player base and the increasing complexity of modern games, has significantly increased global energy consumption, estimated between 230 TWh and 347 TWh annually. The application of generic optimizations when compiling games leads to inefficient performance and unnecessary energy use. Consequently, this work presents a novel method for reducing video games energy consumption by tailoring software optimizations for the specific underlying hardware architecture. This is defined as a combinatorial optimization problem, formulated as finding the optimal sequence of LLVM code transformations that modify the considered game into an optimized version with minimum energy use. A cellular genetic algorithm is employed to optimize Doom Legacy —a modern adaptation of the original Doom engine— running on a Raspberry Pi 4. Greener versions of the software were found, compared to the performance of the popular -O3 compilation flag (the most aggressive generic LLVM optimization), achieving 26 times lower consumption values. This significant improvement results in dozens of joules saved for one hour gameplay, enhancing energy efficiency in consumer electronics as mobile devices and gaming platforms, positively impacting battery life and greenness performance.
PB IEEE
YR 2025
FD 2025-04
LK http://hdl.handle.net/10498/36239
UL http://hdl.handle.net/10498/36239
LA eng
DS Repositorio Institucional de la Universidad de Cádiz
RD 10-nov-2026