Google is making Android even faster, new kernel-specific optimizations

The mobile operating system from Google is poised to make a significant leap in quality in terms of responsiveness and efficiency.

Through the team that manages the Android LLVM toolchain, the California-based company announced the introduction of an advanced technique named Automatic Feedback-Directed Optimization, or more simply AutoFDO.

This update deeply affects the kernel, the core of the software architecture that has the delicate task of managing and coordinating communications between the processor, the individual applications, and all the device’s hardware components.

Android becomes more responsive thanks to new kernel optimizations

Considering that this underlying infrastructure accounts for about 40% of the CPU processing time on smartphones, it is evident that any technical refinement could translate into concrete and immediately perceptible benefits during daily use.

Traditionally, the software compilation phase relies on instructions and rules set in advance. The compiler translates the code written by developers into a language directly understandable by the processor, seeking the utmost efficiency in the arrangement of processes.

However, these choices often stem from theoretical calculations. The solution introduced by AutoFDO consists of overcoming this initial rigidity, enabling the system to learn directly from data derived from users’ habits.

Device tests and code mapping

To gather reliable information useful for this implementation, engineers conducted rigorous simulations in controlled laboratory environments, using Pixel devices for testing. The tests replicated typical interactions with the hundred most popular applications currently available.

During these intense sessions, dedicated diagnostic tools consistently monitored the behavior of the operating system to identify the most heavily used portions of code, defined by practitioners as hot zones.

Recompiling the kernel based on these valuable findings allows the infrastructure to organize its instructions in a much more logical way. The system routines invoked most frequently are processed with higher priority, streamlining the computation operations.

The practical results of this work directly and positively reflect on the smartphone’s final usability. There is indeed a visibly faster startup of applications, smoother performance during scrolling, instant transitions from one open program to another, and a much more careful energy management, which promises to extend battery life noticeably.

When will it arrive on commercial smartphones?

At present, AutoFDO technology is actively being implemented on development branches identified as android16-6.12 and android15-6.6, the versions that form the backbone of the most recent platform iterations, but not yet on Android 17.

In any case, the California-based company’s plans foresee a widespread diffusion of this optimization method also towards all upcoming versions of the base software.

Over time, the designers intend to extend this compilation method well beyond the main core of the system, involving other critical portions of the software.

Among the upcoming elements that will benefit from this treatment are several supplementary system components and, even more relevant for overall performance, hardware drivers developed specifically by individual phone manufacturers to manage peculiar elements such as camera subsystems and network connectivity modems.