32-bit floating point - Revision history

TArntsen: Add 'floating point' to be consistent with 64-bit page

2016-05-25T05:27:32Z

Add 'floating point' to be consistent with 64-bit page

← Older revision		Revision as of 05:27, 25 May 2016
Line 1:		Line 1:
	'''Norsk Data 32-bit single precision format:'''		'''Norsk Data 32-bit single precision floating point format:'''

	The [[NORD-10]] and the [[ND-100]] could be ordered with either of two different hardware floating point formats, the 32-bit format described on this page, or alternatively with [[48-bit floating point]] hardware.		The [[NORD-10]] and the [[ND-100]] could be ordered with either of two different hardware floating point formats, the 32-bit format described on this page, or alternatively with [[48-bit floating point]] hardware.

TArntsen: Rewritten, extended and improved. References added

2016-05-25T04:31:13Z

Rewritten, extended and improved. References added

@@ Line 1: / Line 1: @@
 '''Norsk Data 32-bit single precision format:'''
-'''Special values for exponent:'''
+The [[NORD-10]] and the [[ND-100]] could be ordered with either of two different hardware floating point formats, the 32-bit format described on this page, or alternatively with [[48-bit floating point]] hardware.
 {|
 |-
-| ''All bits zero except sign bit'': || '''Floating underflow'''.
+|bit 31	|| 1 ||Sign bit
 |-
-| ''0 (binary 0, aka 0x00)'': || '''Zero'''. Strictly speaking this is for the ND-500(0). The ND-100 spec says ''floating zero: All 32 bits zero.'' But any ND-100 introduced after the ND-500 would presumably be compatible, if not before.
+|bit 30-22 || 9 ||	Exponent
 |-
 |}
-Mantissa has an implicit 23th bit of 1 (unless mantissa is 0).
+* The mantissa is normalized: 0.5 &le; ''mantissa'' &le; 1
-This is the 0.5 bit, i.e. mantissa is normalised to be 0.5 < mantissa < 1.0
+* The mantissa always has an implicit 23th bit of 1, unless the exponent is zero (see below)
-Decimal point is thus to the left of the implicit 23th bit.
+=== Sources ===

TArntsen: Initial 32-bit floating point page

2013-04-04T14:18:51Z

Initial 32-bit floating point page

New page

'''Norsk Data 32-bit single precision format:'''
{|
|-
| bit 31 || (1 bit) || Sign bit
|-
| bit 30-22 || (9 bits) || Exponent
|-
| bit 21-0 || (22 bits) || Mantissa
|-
|}
Exponent has a bias of +256. (Subtract 256 to get true exponent)

'''Special values for exponent:'''
{|
|-
| ''All bits zero except sign bit'': || '''Floating underflow'''.
|-
| ''0 (binary 0, aka 0x00)'': || '''Zero'''. Strictly speaking this is for the ND-500(0). The ND-100 spec says ''floating zero: All 32 bits zero.'' But any ND-100 introduced after the ND-500 would presumably be compatible, if not before.
|-
|}

Mantissa has an implicit 23th bit of 1 (unless mantissa is 0).
This is the 0.5 bit, i.e. mantissa is normalised to be 0.5 < mantissa < 1.0

Decimal point is thus to the left of the implicit 23th bit.