Investigate efficiency of using `fixed64` over `uint64`

[jules]

I remember reading somewhere before that using a fixed64 in the place of a uint64 would be more efficient, where the number will be big in most cases. This should be investigated, and if true, we should update the types for these numbers.