Limits for floating-point numbers

Non-Confidential

ARM DUI0375H

ARM^® Compiler v5.06 for µVision^® armcc User Guide

Version 5

16.3 Limits for floating-point numbers

This topic describes the characteristics of floating-point numbers.

The following table gives the limits for floating-point numbers. These constants are defined in the float.h include file.

Table 16-3 Floating-point limits

Constant	Meaning	Value
`FLT_MAX`	Maximum value of `float`.	3.40282347e+38F
`FLT_MIN`	Minimum normalized positive floating-point number value of `float`.	1.175494351e–38F
`DBL_MAX`	Maximum value of `double`.	1.79769313486231571e+308
`DBL_MIN`	Minimum normalized positive floating-point number value of `double`.	2.22507385850720138e–308
`LDBL_MAX`	Maximum value of `long double`.	1.79769313486231571e+308
`LDBL_MIN`	Minimum normalized positive floating-point number value of `long double`.	2.22507385850720138e–308
`FLT_MAX_EXP`	Maximum value of base 2 exponent for type `float`.	128
`FLT_MIN_EXP`	Minimum value of base 2 exponent for type `float`.	–125
`DBL_MAX_EXP`	Maximum value of base 2 exponent for type `double`.	1 024
`DBL_MIN_EXP`	Minimum value of base 2 exponent for type `double`.	–1 021
`LDBL_MAX_EXP`	Maximum value of base 2 exponent for type `long double`.	1 024
`LDBL_MIN_EXP`	Minimum value of base 2 exponent for type `long double`.	–1 021
`FLT_MAX_10_EXP`	Maximum value of base 10 exponent for type `float`.	38
`FLT_MIN_10_EXP`	Minimum value of base 10 exponent for type `float`.	–37
`DBL_MAX_10_EXP`	Maximum value of base 10 exponent for type `double`.	308
`DBL_MIN_10_EXP`	Minimum value of base 10 exponent for type `double`.	–307
`LDBL_MAX_10_EXP`	Maximum value of base 10 exponent for type `long double`.	308
`LDBL_MIN_10_EXP`	Minimum value of base 10 exponent for type `long double`.	–307

The following table describes other characteristics of floating-point numbers. These constants are also defined in the float.h include file.

Table 16-4 Other floating-point characteristics

Constant	Meaning	Value
`FLT_RADIX`	Base (radix) of the ARM floating-point number representation.	2
`FLT_ROUNDS`	Rounding mode for floating-point numbers.	(nearest) 1
`FLT_DIG`	Decimal digits of precision for `float`.	6
`DBL_DIG`	Decimal digits of precision for `double`.	15
`LDBL_DIG`	Decimal digits of precision for `long double`.	15
`FLT_MANT_DIG`	Binary digits of precision for type `float`.	24
`DBL_MANT_DIG`	Binary digits of precision for type `double`.	53
`LDBL_MANT_DIG`	Binary digits of precision for type `long double`.	53
`FLT_EPSILON`	Smallest positive value of x that 1.0 + x != 1.0 for type `float`.	1.19209290e–7F
`DBL_EPSILON`	Smallest positive value of x that 1.0 + x != 1.0 for type `double`.	2.2204460492503131e–16
`LDBL_EPSILON`	Smallest positive value of x that 1.0 + x != 1.0 for type `long double`.	2.2204460492503131e–16L

When a floating-point number is converted to a shorter floating-point number, it is rounded to the nearest representable number.
Floating-point arithmetic conforms to IEEE 754.

Non-Confidential	PDF version	ARM DUI0375H
Copyright © 2007, 2008, 2011, 2012, 2014-2016 ARM. All rights reserved.