Interactive CPython Stack & Heap Memory Explorer

Welcome to the CPython Memory Architecture Interactive Lab.

High-performance AI pipelines in Python frequently suffer from silent memory bloat, unexpected object copies, and uncoordinated garbage collection pauses. In this hands-on laboratory, you will directly interact with CPython’s memory structures.

Lab Objectives

1. Trace pointer allocations and reference count increments (ob_refcnt) during variable binding.
2. Observe how container types (list, dict, tuple) create cyclic references that bypass immediate refcount deallocation.
3. Test the CPython generational garbage collector thresholds (Generations 0, 1, and 2).

Interactive Memory Simulator

Use the visual memory explorer below to trigger pointer allocations, simulate variable scope exits, and inspect how Python reclaims heap memory in real-time.

import sys
import gc

class Node:
    def __init__(self, val):
        self.val = val
        self.next = None

# 1. Create a cyclic reference
a = Node("Head")
b = Node("Tail")
a.next = b
b.next = a  # Cycle created!

print(f"Node A refcount: {sys.getrefcount(a) - 1}")
print(f"Node B refcount: {sys.getrefcount(b) - 1}")

# 2. Break scope references
del a
del b

# At this point, the cycle is unreachable by user code but still in memory!
unreachable = gc.collect()
print(f"Unreachable circular objects reclaimed by GC: {unreachable}")

Key Engineering Takeaways

• Small Object Allocator (pymalloc): CPython bypasses malloc() for allocations $\le 512$ bytes, reserving arenas (256 KB) subdivided into pools (4 KB) and blocks.
• Weak References: Use import weakref when building caches or graph neural network adjacency graphs to prevent cyclic reference memory leaks without requiring constant gc.collect() overhead.