Zygote art and app startup

Zygote ART and App Startup Deep Dive

Overview

Android startup performance combines process forking via Zygote and runtime execution strategy in ART (AOT + JIT + profile-guided optimization).

Why Zygote exists

Zygote preloads common framework classes/resources once, then forks apps. Copy-on-write allows sharing clean pages until mutation.

Benefits:

• faster process creation
• reduced duplicated memory
• lower startup initialization cost

ART execution strategy

ART balances install cost and run-time speed with:

• baseline profile guided compile
• JIT for runtime hotspots
• selective AOT artifacts

Startup hidden costs

• class verification and linking at first touch
• heavy static initializers
• deep dependency graph pull-in

Move expensive setup off critical first-frame path when possible.

Baseline profiles

Profiles mark methods that should be precompiled for startup and key journeys. Stale profiles regress first-run performance after feature changes.

Measurement

Track:

• cold/warm/hot startup times
• first frame and first interaction latency
• class loading and initialization events in trace

Interview guidance

Good answers discuss both memory and startup speed. Great answers include an operational loop: generate profile, ship, measure, refresh.