How the hash function is implemented for `object`

Asked Dec 15 '20 at 08:02

Active Dec 15 '20 at 08:15

Viewed 34 times

The comparison of the following benchmark results in 0.1 sec vs 0.12 sec on my computer. I first thought the result is strange. I thought __hash__ will be applied to all the fields of the class, therefore, the hashing time should be equal.

However, afterward, I notice that object class has its own __hash__ function, from the documentation.

I want to know the reason why hashing object is that fast. More specifically, I want to know how the __hash__ function is implemented for the object class. Is object.__hash__ compute the hash of pointer of the object internally?

import time

string = "hash_test"
class A(object):
    def __init__(self, string):
        self.a = (string, string, string, string)
a = A(string)

ts = time.time()
for i in range(1000000):
    hash(a)
print(time.time() - ts) # 0.1 sec

ts = time.time()
str_tuple = (string, string, string, string)
for i in range(1000000):
    hash(str_tuple)
print(time.time() - ts) # 0.12 sec

edited Dec 15 '20 at 08:12

asked Dec 15 '20 at 08:02

orematasaburo

1,207
10
20

https://stackoverflow.com/questions/12020984/hash-table-why-is-it-faster-than-arrays#:~:text=Hash%20tables%20are%20a%20bit,key%2C%20you%20compute%20the%20hash. – Huzefa Sadikot Dec 15 '20 at 08:05
your timing is biased. you should define `sl = (string, string, string, string)` first then clock `hash(sl)` – Jean-François Fabre Dec 15 '20 at 08:07
@Jean-FrançoisFabre true – orematasaburo Dec 15 '20 at 08:09
the hash doesn't take the array in the case of the object – Jean-François Fabre Dec 15 '20 at 08:10
But, in general the [data model docs is where you look for this kind of stuff](https://docs.python.org/3/reference/datamodel.html#object.__hash__) – juanpa.arrivillaga Dec 15 '20 at 08:21

How the hash function is implemented for `object`

0 Answers0