Android Runtime - How Dalvik and ART Work

Source

• Type: webpage
• Origin: https://proandroiddev.com/android-runtime-how-dalvik-and-art-work-6e57cf1c50e5
• Author: Paulina Sadowska
• Imported: 2026-04-20

Content

This article explains how Android Runtime translates app bytecode (.dex) into CPU-executable machine code, and how compilation strategies evolved from Dalvik to ART to balance performance, RAM usage, and install/update cost.

Runtime basics

• APKs include .dex bytecode for app/library logic.
• Android Runtime compiles this bytecode to machine code at runtime or ahead of runtime depending on strategy.
• Runtime behavior also includes memory management and garbage collection (not the focus of this article).

Dalvik era (up to Android K)

• Dalvik emphasized low RAM usage on early devices.
• It relied on JIT (Just-In-Time) compilation: compiling code during execution.
• Benefit: lower memory pressure from not precompiling everything.
• Trade-off: slower runtime performance due to on-the-fly compilation.

ART in Android L

• ART introduced AOT (Ahead-Of-Time) compilation.
• Code is compiled before execution, improving app runtime speed significantly.
• Trade-offs:
  • higher storage/RAM cost,
  • slower app installation and system updates due to broad precompilation.

Android N: profile-guided hybrid model

• Android Runtime reintroduced JIT plus profile-guided compilation.
• Frequently executed ("hot") methods are identified and precompiled/cached.
• This hybrid approach improves hot-path performance while avoiding full precompile overhead.
• Precompilation is scheduled during idle/charging windows to reduce user impact.

Android P: cloud/common profiles

• Google introduced cloud-distributed common profiles for new installs.
• A "common core profile" precompiles methods/classes commonly used by most users.
• Device-specific profiling still continues after install for user-specific optimization.

Key Takeaways

• Dalvik prioritized RAM efficiency with JIT; ART initially prioritized execution speed with AOT.
• Modern Android Runtime uses a hybrid JIT + profile-guided model for better real-world efficiency.
• Cloud/common profiles improve first-run experience by pre-optimizing frequently used code paths.
• Runtime compilation strategy is a major factor in startup speed, install cost, and overall UX.