Linear Assembly For Full Live-Too-Long Code - Texas Instruments TMS320C6000 Programmer's Manual

Example 6–58. Linear Assembly for Full Live-Too-Long Code
.global _live_long
_live_long: .cproc
.reg ai, bi, sum0, sum1, sum
.reg a0p, a_0, a_1, a_2, a_3, b_0, b0p, b_1, b_2, b_3, cntr
MVK 100,cntr
ZERO sum0
ZERO sum1
LOOP: .trip 100
LDH .D1
LDH .D2
MPY .M1
SHR .S1
MPY .M1X
MV .D1
ADD .L1
ADD .L1
MPY .M2X
SHR .S2
MPY .M2X
MV .D2
ADD .L2
ADD .L2
[cntr] SUB .S2
[cntr] B .S1
ADD sum0,sum1,sum
.return sum
.endproc
a, b, c, d, e
*a++,ai
*b++,bi
ai,c,a_0
a_0,15,a_1
a_1,d,a_2
a_0,a0p
a_2,a0p,a_3
sum0,a_3,sum0
bi,c,b_0
b_0,15,b_1
b_1,e,b_2
b_0,b0p
b_2,b0p,b_3
sum1,b_3,sum1
cntr,1,cntr
LOOP
Optimizing Assembly Code via Linear Assembly
; cntr = 100
; sum0 = 0
; sum1 = 0
; load ai from memory
; load bi from memory
; a0 = ai * c
; a1 = a0 >> 15
; a2 = a1 * d
; save a0 across iterations
; a3 = a2 + a0
; sum0 += a3
; b0 = bi * ci
; b1 = b0 >> 15
; b2 = b1 * e
; save b0 across iterations
; b3 = b2 + b0
; sum1 += b3
; decrement loop counter
; branch to loop
; Add sumi and sumi+1 for ret value
Live-Too-Long Issues
6-107