为什么一个 numpy 数组有 96 字节的开销?
Why does a numpy array have 96 bytes of overhead?
如果我采用一个简单的空 numpy 数组,我可以看到它有 96 个字节的开销,
>>> sys.getsizeof( np.array([]) )
96
那96字节存储的是什么? numpy 或 Python 3 (cpython) 的 C 源代码中的何处设置?
数组存在于 numpy/core/include/numpy/ndarraytypes.h
的 C 源代码中
参见:https://github.com/numpy/numpy/blob/master/numpy/core/include/numpy/ndarraytypes.h
看起来它有几个指针、维数和 PyObject_HEAD,它们加起来可能会达到您看到的字节数。
/*
* The main array object structure.
*/
/* This struct will be moved to a private header in a future release */
typedef struct tagPyArrayObject_fields {
PyObject_HEAD
/* Pointer to the raw data buffer */
char *data;
/* The number of dimensions, also called 'ndim' */
int nd;
/* The size in each dimension, also called 'shape' */
npy_intp *dimensions;
/*
* Number of bytes to jump to get to the
* next element in each dimension
*/
npy_intp *strides;
PyObject *base;
/* Pointer to type structure */
PyArray_Descr *descr;
/* Flags describing array -- see below */
int flags;
/* For weak references */
PyObject *weakreflist;
} PyArrayObject_fields;
如果我采用一个简单的空 numpy 数组,我可以看到它有 96 个字节的开销,
>>> sys.getsizeof( np.array([]) )
96
那96字节存储的是什么? numpy 或 Python 3 (cpython) 的 C 源代码中的何处设置?
数组存在于 numpy/core/include/numpy/ndarraytypes.h
的 C 源代码中参见:https://github.com/numpy/numpy/blob/master/numpy/core/include/numpy/ndarraytypes.h
看起来它有几个指针、维数和 PyObject_HEAD,它们加起来可能会达到您看到的字节数。
/*
* The main array object structure.
*/
/* This struct will be moved to a private header in a future release */
typedef struct tagPyArrayObject_fields {
PyObject_HEAD
/* Pointer to the raw data buffer */
char *data;
/* The number of dimensions, also called 'ndim' */
int nd;
/* The size in each dimension, also called 'shape' */
npy_intp *dimensions;
/*
* Number of bytes to jump to get to the
* next element in each dimension
*/
npy_intp *strides;
PyObject *base;
/* Pointer to type structure */
PyArray_Descr *descr;
/* Flags describing array -- see below */
int flags;
/* For weak references */
PyObject *weakreflist;
} PyArrayObject_fields;