CompuServe Messages

There are, in the PC architecture, 3 types of floating point numbers. The difference is only on the amount of precision. float (or single precision) are 32 bits in length from which 8 bits are used for the Exponent and 23 for the Mantissa (1 for sign). It can represent a number from 1.18×10^-30 (1.18E-30) to 3.40×10^38. That's 30 decimal places for a negative number and 38 for a positive number. double (or double precision) are 64 bits in length from which 11 bits are used for the Exponent and 52 for the Mantissa. It can represent a number from 2.23×10^-308 to 1.80×10^308 (308 decimal places). long double (or neurotic precision) are 80 bits in length from which 15 bits are used for the Exponent and 64 for the Mantissa. It can represent a number from 3.36×10^-4932 to 1.19×10^4932 (4,932 decimal places). Errors in conversion are accumulative but I doubt it very much it would make any difference even with single precision. With a double precision you could split an atom at the edge of the solar system (provided you had what to aim with). — What are the cons of double-precision? Basically that it forces the system to use twice the amount of memory for all geometry. Eventually, when RAM is cheap enough, this won't be an issue. But for now… – G