@736939

6:08 Does it use vectorization, under the hood, when we do chaining? Or how can we achieve the performance of "numpy"? Thank you.

@AK-vx4dy

Excelent work as in other part!
@9:12 Is there error on slide ? in green box you have "sum += nums" instead of "sum += n"

@karthik947

@00:18 - Tell me you're german without telling me you are german :)

@nejathakan5521

Excellent

@saaddahmani1870

Very good. Keep going please.

@rcoder01

Would you be able to share your testing methodology for the performance of the three kinds of loops? Copying an i32 should be extremely cheap, and should be faster than dereferencing a &i32, which you would assume to be happening with the `&nums` loop and the for and while loops. I tested the four kinds of loops you showed and found that the `for n in nums` loop was consistently faster than the `for n in &nums` loop (as I would expect) and the `for` and `while` loops I wasn't able to get consistent data on, but both were generally slower than `for n in nums`.

Saying that copying is "very very costly" is quite disingenuous because copying is only expensive for large data types, but an i32 is only 4 bytes long whereas a &i32 is actually 8 bytes long on a 64-bit architecture, which is more expensive to hold in RAM and could lead to more cache misses in some circumstances.

@alessandrodarcangeli9005

Hi, i would like to know here its possible check that the first loop make copies?, thank you in advance