Resident Evil 7 was successfully launched on a OnePlus 12R with Snapdragon 8 Gen 2 at a stable 30 FPS thanks to AI, which helped identify emulation issues and optimize game data.
Initially, the smartphone produced about 20 FPS. After connecting to a Mac and enabling debug mode, the Hermes agent, based on DeepSeek-V4.1-Flash, analyzed the game’s performance and identified limitations related to I/O and memory bandwidth.
AI suggested compressing textures from 4096 to 1024 and 512 pixels. The volume of necessary data decreased from approximately 25 to 8 GB, and after removing unnecessary files, it amounted to about 13 GB.
Additionally, Snapdragon Game Super Resolution was used: the game was rendered at 720p and scaled to approximately 1200p. As a result, Resident Evil 7 ran at a stable 30 FPS, consuming about 5–6 W. At the same time, the smartphone was simultaneously recording gameplay, creating additional load.
The experiment demonstrates the potential of AI for optimizing emulated games by identifying specific problems and adjusting data and rendering.
