Coordinating Space and Time
summary
Shows a concrete pixel counter example that traces registers, L1/L2 cache and 4 GB RAM and a quad core phone, then demonstrates an initialized boolean reordering race that can crash or silently lose data. Reproduces a Google Android sample bug where concurrent LinkedHashMap.put corrupts state. Walks through volatile, synchronized, atomics, executors, concurrent collections, blocking queues and RxJava and ends with a hard rule: make models on the main thread and prefer immutability.
AI-written summary · about
chapters (8)
- 0:00Introduction to concurrency
- 5:03CPU and RAM interactions
- 8:28Naive vs optimized code
- 17:06Memory hierarchy and caches
- 22:19The dangers of reordering
- 26:41Real-world concurrency bugs
- 31:00Concurrency tools and APIs
- 38:40Effective concurrency strategy
related talks
42mAdvanced techniques for concurrency & memory management (en)
41mEmbracing the Chaos: Investigating Production Madness
08h 39mLivestream | Day 2, theater 2
45mHigh performance in Android
49mKeynote: Android Development
42mSpeaking of Leaking
45mExploring the possibilities of Unidirectional Data Flow Architectures on Android (en)
40mAndroid high performance baby