3 ms·
Interesting, but this only admits 32-bit floats. The original idea behind NaN-tagging is to overlap pointer representation with 64-bit floats.
by tomp 4y ago
Interesting, but this only admits 32-bit floats.
The original idea behind NaN-tagging is to overlap pointer representation with 64-bit floats.
- moonchild 4y ago> The original idea behind NaN-tagging is to overlap pointer representation with 64-bit floats. And that's why this is 'another NaN-based tagging strategy'.