I have a tuple T, consisting of

1. An integer id in range [1, 70 × 10⁶]
2. A double score₁ in range [0.0, 1.0]
3. A double score₂ in range [0.0, 1.0]
4. An integer x in range [0, 2]

What can be the most efficient way to store such tuple T =  < id, score₁, score₂, x >  given that only 5 digits after the decimal are sufficient for score₁ and score₂.

Currently, I am storing this tuple as a long which takes 8 bytes of memory. 27 bits for id, 17 bits for score₁, 17 bits for score₂ (considering values upto 5 decimal digits), and 2 bits for x, Overall 63 bits.

I'm trying to optimize the memory usage as there are millions of such tuples in an array. Also, If your way can store digits more than 5 decimal places, please tell me.

Is there a way, by which we can store the tuple T using float? given that if the tuple is hashed to float, I should be able to get back id, x accurately and score₁, score₂ without much errors.

Thanks for your time, have a nice day.