The bytecode gobbledygook
summary
Bytecode hides costly choices: javac output forces boxing, iterator overhead and repeated allocations that trash GC and slow loops; replacing wrappers with primitives and using arrays instead of for-each often cuts time dramatically — measured up to 18% faster while JITting and seen roughly 2x penalty on for‑each/Iterator on a Nexus 5. Tool-specific facts matter: Jack and Jill deprecated in favour of D8 compiler and Java 8 tooling. Measure assembly dumps and timed runs before changing code.
AI-written summary · about
chapters (16)
- 0:00Start
- 6:38Imagine this simple C code
- 6:53GCC compiler
- 7:29Let's do a small change
- 8:55JIT assembly output
- 9:17Autoboxing
- 15:34Loops - List
- 17:10Loops - Array
- 18:02Loops - size cached
- 23:03Manual bytecode optimization
- 24:47Worth it?
- 25:23Overhead of calling a method
- 26:05String concatenation
- 27:43String.concat()
- 35:02Tooling Disassembler - ART
- 35:31Tooling PrintAssembly - JIT
