[sundarurfriend]

1000_0000 is 10 million, you want 1000_000. And k is better initialized as `Int32(1):Int32(1000_000)`, which is significantly faster (though it doesn't matter for the measurement here).

Checking with a million values, on my Julia v1.8.0, the dict takes up 18 MB, or about 130% more than the plain values (or an array of Pairs) would.

[kwertzzz]

Thank you for correcting this! (I get the same numbers as you for a million elements).

[adgjlsfhk1]

note that the overhead will vary a ton based on the number of elements you use. I believe Julia's dicts currently resize to be 4x current size (to avoid hash collisions), so you should see anywhere from 30% extra to 300% extra depending on how many elements you have. There has been some effort recently to move to a more SwissDict-like approach which should reduce the memory overhead.