[Python] Why the performance difference between `|` and `or` different for different values?

I was timing performance for operators, I found this:

>>> timeit.timeit("10000000<2 or 10000000>1000",number=10000000)
0.4213107880004827
>>> timeit.timeit("(10000000<2) | (10000000>1000)",number=10000000)
0.5472174039969104

>>> timeit.timeit("1000<200000 or 100>1000",number=10000000)
0.26014828799998213
>>> timeit.timeit("(1000<200000) | (100>1000)",number=10000000)
0.49938497000039206


Although the difference is sometimes miniscule, or consistently beats | for these large number comparisons (I am not really sure if this is because of the large numbers or not).

But then:

>>> timeit.timeit("0 or 1",number=10000000)
Out[134]: 0.17863203100023384
>>> timeit.timeit("0 | 1",number=10000000)
Out[135]: 0.06668136799999047

>>> timeit.timeit("00000 | 111111",number=10000000)
Out[136]: 0.06739834499967401
>>> timeit.timeit("00000 or 111111",number=10000000)
Out[137]: 0.1818816989998595

>>> timeit.timeit("True or False",number=10000000)
Out[138]: 0.12497510200046236
>>> timeit.timeit("True | False",number=10000000)
Out[139]: 0.06938139100020635

>>> timeit.timeit("False | True",number=10000000)
Out[140]: 0.0667262490005669
>>> timeit.timeit("False or True",number=10000000)
Out[141]: 0.1443343150003784


What's happening? Is this expected? I believed that | always checks both operands even if the first one evaluates to True. So at least for cases where 1st operand is True, shouldn't or be consistently faster?

Continue reading...