5 ms·
Would a "fixed-point binary-coded decimal" type be a solution here? With 64 bit values that gives you 16 digits to play with, which for "everyday numbers" seems
by mysterydip 4y ago
Would a "fixed-point binary-coded decimal" type be a solution here? With 64 bit values that gives you 16 digits to play with, which for "everyday numbers" seems like plenty.
- gpderetta 4y agoIEEE double gives you 15 digits and a much larger dynamic range, so the tradeoff is just not worth it except for specialized applications.