Texas Instruments TMS320C3x 사용자 설명서
Floating-Point Conversion (IEEE Std. 754)
5-18
Example 5–4.IEEE-to-TMS320C3x Conversion (Fast Version) (Continued)
*
NOTE:
SINCE THE STACK POINTER SP IS USED, MAKE SURE TO
*
INITIALIZE IT IN THE CALLING PROGRAM.
*
*
*
*
CYCLES: 12 (WORST CASE) WORDS: 12
*
.global FMIEEE
*
FMIEEE
AND3
R0,*AR1,R1
; Replace fraction with 0
BND
NEG
; Test sign
ADDI
R0,R1
; Shift sign
; and exponent inserting 0
; and exponent inserting 0
LDIZ
*+AR1(1),R1
; If all 0, generate C30 0
SUBI
*+AR1(2),R1
; Unbias exponent
PUSH
R1
POPF
R0
; Load this as a flt. pt. number
RETS
*
NEG
NEG
PUSH
R1
POPF
R0
; Load this as a flt. pt. number
NEGF
R0,R0
; Negate if orig. sign is negative
RETS
Example 5–5 shows the complete conversion between IEEE and ’C3x formats.
In addition to the general case and the 0s, it handles the special cases as follows:
In addition to the general case and the 0s, it handles the special cases as follows:
-
If NaN (e = 255, f < > 0), the number is returned intact.
-
If infinity (e = 255, f = 0), the output is saturated to the most positive or nega-
tive number, respectively.
tive number, respectively.
-
If denormalized (e = 0, f< >0), two cases are considered. If the MSB of f is
1, the number is converted to ’C3x format. Otherwise an underflow occurs,
and the number is set to 0.
1, the number is converted to ’C3x format. Otherwise an underflow occurs,
and the number is set to 0.