Verified Answer :

'm pretty sure the big O is actually O(log n)

Not really a proof but anyway.

If you write out the values before each iteration of the loop you get.

L - iteration number

L i t
1 1 0
2 1 2
3 3 6
4 9 18
5 27 54
6 81 162

So if n was 80 it would take 6 iterations

You can see i is growing like 3^(L-2)

So if n was 1000

1000 = 3^(L-2)
log(1000) [base 3] = L - 2
log(1000) [base 3] + 2 = L
8.287709822868152 = L

and if you ceil the answer you get L = 9

and just to check with the table
6 81 162
7 243 486
8 729 1458
9 2187 ...

The table isn't perfect but hopefully you get the main idea.