Memory Management

Memory management in Python involves a private heap containing all Python objects and data structures. The management of this private heap is ensured internally by the Python memory manager

On top of the raw memory allocator, several object-specific allocators operate on the same heap and implement distinct memory management policies adapted to the peculiarities of every object type

Allocator Domains

All allocating functions belong to one of three different “domains”

• Raw domain: intended for allocating memory for general-purpose memory buffers where the allocation must go to the system allocator or where the allocator can operate without the GIL. The memory is requested directly to the system.
• “Mem” domain: intended for allocating memory for Python buffers and general-purpose memory buffers where the allocation must be performed with the GIL held. The memory is taken from the Python private heap.
• Object domain: intended for allocating memory belonging to Python objects. The memory is taken from the Python private heap.

Raw Memory Interface

• void *PyMem_RawMalloc(size_t n)
• void *PyMem_RawCalloc(size_t nelem, size_t elsize)
• void *PyMem_RawRealloc(void *p, size_t n)
• void PyMem_RawFree(void *p)

Memory Interface

• void *PyMem_Malloc(size_t n)
• void *PyMem_Calloc(size_t nelem, size_t elsize)
• void *PyMem_Realloc(void *p, size_t n)
• void PyMem_Free(void *p)
• PyMem_New(TYPE, n)
• PyMem_Resize(p, TYPE, n)
• void PyMem_Del(void *p)

Object allocators

• void *PyObject_Malloc(size_t n)
• void *PyObject_Calloc(size_t nelem, size_t elsize)
• void *PyObject_Realloc(void *p, size_t n)
• void PyObject_Free(void *p)

The pymalloc allocator

Python has a pymalloc allocator optimized for small objects (smaller or equal to 512 bytes) with a short lifetime. It uses memory mappings called “arenas” with a fixed size of 256 KiB. It falls back to PyMem_RawMalloc() and PyMem_RawRealloc() for allocations larger than 512 bytes.