A simple loop is faster in python, adding types makes it 4-8 times slower

[mahmoudajawad]

Take the following mojo block:

from Time import now

mojo_t1 = now()
for i in range(10000):
    pass
mojo_t2 = now()

print(mojo_t2 - mojo_t1)

I get between 250 to 320 nanoseconds.

Same in Python doesn't go beyond 260 nanoseconds.

What is more interesting is re-writing it to:

from Time import now

let mojo_t1: Int = now()
for i in range(10000):
    pass
let mojo_t2: Int = now()

print(mojo_t2 - mojo_t1)

take the execution time to 1550-2150.

What is going on here?

[Mogball]

The execution time you are measuring is entirely the notebook backend executing the code line-by-line. The loop doesn't execute at all -- it's optimized away since the body is empty. The time you are measuring is likely noise. In the first snippet, the type of mojo_t1 is already going to be Int, since it is inferred from the result type of now().