Google introduces AutoFDO to optimize Android kernel based on real usage, improving app launch speeds, smoothness, and battery life in current and future versions.
© RusPhotoBank
Google is introducing a new optimization technology into the Android kernel—the core part of the operating system that manages interactions between the processor, applications, and hardware. This method, called Automatic Feedback-Directed Optimization (AutoFDO), enables the system to learn from real device usage and reconfigure code for more efficient performance.
The Android kernel plays a crucial role in overall performance. According to Google, it accounts for about 40% of a smartphone's total CPU time. Even minor improvements at this level can significantly impact app launch speeds, interface smoothness, and energy consumption. Previously, the compiler optimized code based on general algorithms and assumptions, without considering actual usage scenarios.
Now, Google collects data in lab settings using Pixel smartphones. During testing, the system runs and actively uses the hundred most popular apps, while profiling tools identify which parts of the kernel are used most frequently. These hot spots then receive priority optimization when the kernel is rebuilt.
This allows the compiler to organize code more rationally, speeding up the execution of the most in-demand processes. As a result, users may notice faster app launches, smoother multitasking, and potentially longer battery life.
The technology is already being implemented in the android16-6.12 and android15-6.6 branches, which are used in current system versions. Going forward, Google plans to extend AutoFDO to additional kernel components and hardware drivers, including camera and modem modules. This move could deliver even greater performance gains in future Android releases.