Intel 80287, 80286 manual It116

NUMERIC PROGRAMMING EXAMPLES

iAPX2Sb MACRO ASSEMBLER 80287 Floating-Point to IS-Digit ASCII Conv~1"6ion 10: 12: 38 09/25/83 PAGE

LOC aSJ LINE SOURCE

51

52

53

54

55

56

57

'8

59

If the given number Ula!> zero, the ASCII string will contain a sign

and a single zero chacter. The value string_size indicates. the total

length of the ASCII string including the sign character. StringeD) will

alwal,ls hold the sign. It is possible for string_sile to be less than

field_sileo This occurs for zeroes or integer values. A pauedo zero

will return a special return code. The denormal count will indicate

the power of two original III associated with the vlillue. The power of ten and ASCII string will be as if the value was an o'f'dinar~ lero.

60This subroutine is accurate up to a maximum of 18 decimal diilits for

61integers. Integer values will have a decimal power of zero associated

62with them. For non integers. the result will be accurate to within 2

63dIJcimal digits of the 16th decimal place (double precision>. The

64exponentiate instruction is also used for scaling thIJ value into the

65range acceptable for the BCD data t~pe. The rounding mode in effect

66on entrv to the subroutine is used for thIJ conversion.

	67
	68		The following regist.~rs are not transpart'nt:
	69
	70
	71
	72	+1
	73	,	Dt'fine the stack lallout.
	74
	75			'0·
OOOOC]	76	bp_save			word ptr [bpJ
0002t:]	77	es_save			bp_save + size bp_save
				'0·
					es_save + size es_save
0004[1	78	returnJtr		'0·
0006[]	79	power-ptr		'0·	return_ptr + size returnJltr
OOOSt]	80	field_size		'0·	powerJltr + sile power_ptr
CODAr]	81	SileJltr		'0·	field_size + size field_sil,e
oOOetl	82	string-ptr		'0·	size_ptr + size sizeJltr
OOOEt]	83	denormal_ptr		'0·	string_ptr + size stringJtr
	84			'0·
	8.	parms_sl ze			size power_ptr + !;ize field_size + size size-ptr +
OOOA	87	•			size stringJltr + size denormalJtT·
	86
	88	,	Define	constants		used
0012	89			'0·		, Number of iigits in bcd_value
	90	BCD_DIGITS			18
0002	91	WORD_SIZE			2
				eo·
OOOA	92	BCD_SIZE		'0·	10	, Define return values
0001	93	MINUS		eo·	I
0004	94	NAN		'0·	4	, The exact values chosen here are
0006	95	INFINITY		'0.	6	important. The\! must correspond to
0003	.6	INDEFINITE		'0·	3	, the possible return v .. lu •• and b. in
0008	97	PSUEDO_ZERO		'0·	-
					8	the same numeric order as tested b~
-0002	98	INVALID		'0·	2	the program.
-0004	9.	ZERO		'0·	-6.
-0006	100	DENORMAL
-oooa	101	UNNORMAL		'0·	-8
				'0·
0000	102	NORMAL		'0·	0
0002	103	EXACT		'0·	2
	104	,
	105		Define la~out of temporOlr\! storage area.
-OOOiii![]	106			.,.
	107	status			word ptr tbp-WORD_SIZEl
-0004C]	108	power _two		'0·	status - WORD_SIZE
-COObC]	109	power _ten		'0·	power_two - WORD_SIZE
-OOIOC]	110	bcd_value		'0·	tbyte ptr power _ten - BCD_SIZE
-OOIOC]	III	bcd_byte		eo·	bl,lte ptr bcd_value
-OOIOC]	112	fr.c t i on		'0·	bcd_value
	113
	114				size status + size power_two + 6ize power_ten
0010	It116.				+ size bcd_v.lue
	lt7
	118	+1
	119			segment	e1"	pub I ic
	120			ext1"n	power_tab I.: q,ward
	121

		12.
		123
0000 DADO		124
		12.
		126
		127
		128
0002 FB		129
		130
0003	04
0004	F9
uuv;;;	u;;;
0006	00	131
0007	06
0008	01 '
0009	07
OOOA Fe		132
0009 FE
Dooe FD
0000	FE
DOOE	FA	133
DOOF	FE
0010	FB
0011	FE

Constant. used blJ this function.

con.tlD	even		j	Optimize for 16 bits
	d.	10	J	AdJustment value floT"	too big BCD

Convert the C3.C2.Cl.CO encoding fir om tos_stOltus into meaningful bit fllags OInd values.

dbUNNORMAL. NAN. VNNORMAl. + MINUS. NAN + MINUS.

NORMAL. INFINITY. NORMAL + MINUS. INFINITY + MINUS.

ZERO, INVAl.lD. ZERO + MINUS, INVALID,

DENORMAL, INVALID, OENORMAL + MINUS. INVALID

Figure 4-6. Floating-Point to ASCII Conversion Routine (Cont'd.)

4-8