额外的数据会降低Python对象的性能吗？

Question

I could probably run experiments to measure this myself, but I'm hoping somebody can provide a detailed answer.

Suppose I have two Python arrays:

small_array = [{"x": i, "foo": "hamburger"} for i in range(0,10000)]
big_array = [{"x": i, "foo": "hamburger", "bar": "hotdog"} for i in range(0,10000)]

My question is: will basic array operations (such as iteration or accessing by index) that involve only the "x" parameter be faster for small_array compared to big_array ?

I'm asking because I often find myself building a complex data structure X on which I will perform expensive operations A and B such that the overlap between the attributes of X used by A and B is small. So I'm wondering if there are performance advantages associated with separating X into Y and Z so that A can operate on Y and B can operate on Z.

Answer 1

每个列表的迭代和索引编制速度相同（对于以后的Google搜索，它们是列表，而不是数组）。

Answer 2

The performance difference on reasonably-sized (1000 or fewer items) dicts is statistically insignificant. There may be some degenerate cases where many of the keys hash to the same value but the hash randomization found in modern versions of Python sweeps this under the rug.

As for the list itself, each element is only a reference to another object so the size of the underlying object does not affect list operations.