www.digitalmars.com         C & C++   DMDScript  

c++ - How many bits in the mantissa if I declare all floats as long doubles ?

reply Nicholas Jordan <Nicholas_member pathlink.com> writes:
The source code for an entrophy testing method I am trying to incorporate in my
program has the comment:

/* Bytes used as Monte Carlo co-ordinates.
This should be no more bits than the mantissa
of your "double" floating point type. */

#define MONTEN	6

I cannot determine from either the .asm or the .h or the help files how many
bits are in the mantissa.
Jun 18 2006
parent reply Walter Bright <newshound digitalmars.com> writes:
Nicholas Jordan wrote:
 The source code for an entrophy testing method I am trying to incorporate in my
 program has the comment:
 
 /* Bytes used as Monte Carlo co-ordinates.
 This should be no more bits than the mantissa
 of your "double" floating point type. */
 
 #define MONTEN	6
 
 I cannot determine from either the .asm or the .h or the help files how many
 bits are in the mantissa.

long doubles have a 64 bit mantissa, a 15 bit exponent, and a 1 bit sign.
Jun 18 2006
parent Scott Michel <scottm aero.org> writes:
Walter Bright wrote:
 Nicholas Jordan wrote:
 The source code for an entrophy testing method I am trying to
 incorporate in my
 program has the comment:

 /* Bytes used as Monte Carlo co-ordinates.
 This should be no more bits than the mantissa
 of your "double" floating point type. */

 #define MONTEN    6

 I cannot determine from either the .asm or the .h or the help files
 how many
 bits are in the mantissa.

long doubles have a 64 bit mantissa, a 15 bit exponent, and a 1 bit sign.

I'd really like to see a SEF float format with two or more sign bits. :-)
Jun 19 2006