Interpreter Internals & Executable Architecture

Introduction

Python source code is never executed directly by the CPU. Every .py file passes through a compiler that produces platform-independent bytecode, which is then executed by the Python Virtual Machine (PVM). Understanding this three-stage pipeline (source -> bytecode -> machine execution) is crucial for diagnosing performance issues and selecting the right runtime.

Key Concepts

• ›CPython: the reference interpreter. Compiles to .pyc bytecode and evaluates in a C loop.
• ›JIT (Just-In-Time): PyPy traces hot execution paths and compiles them to native machine code for large speedups.
• ›Bytecode is cached in __pycache__/ directories; invalidated when source mtime/version changes.
• ›The DIS module allows inspecting the produced bytecode.

Syntax & Usage

Use dis.dis() to inspect a function's bytecode. Choose CPython for ecosystem compatibility, PyPy for long-running CPU-bound numeric loops.

Example

def greet(name): return 'Hello ' + name

Importing and disassembling reveals LOAD_GLOBAL / LOAD_CONST / RETURN_VALUE opcodes.

Common Mistakes

• ›Assuming Python compiles to native machine code
• ›Expecting PyPy to speed up C-extension libraries (they may actually be slower)
• ›Ignoring the import-time cost of top-level statements

Practice

Describe an interpreter pipeline and identify where bytecode caching occurs.

Key Takeaways

• ›Python always compiles source to bytecode first
• ›PyPy's JIT trades warm-up time for steady-state speed
• ›Bytecode is portable across platforms, not rebindable across Python versions