Texas Instruments TMS320C6000 Programmer's Manual page 186

Example 6–29. Assembly Code for Floating-Point Dot Product (Software Pipelined
With No Extraneous Loads) (Continued
LOOP:
LDDW
|| LDDW
|| MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
||[A1] B
||[A1] SUB
; Branch occurs here
MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
MPYSP
|| MPYSP
|| ADDSP
|| ADDSP
ADDSP
|| ADDSP
ADDSP
|| ADDSP
ADDSP
|| ADDSP
ADDSP
|| ADDSP
.D1 A4++,A7:A6
.D2 B4++,B7:B6
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.S2 LOOP
.S1 A1,1,A1
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.M1X A6,B6,A5
.M2X A7,B7,B5
.L1 A5,A8,A8
.L2 B5,B8,B8
.L1 A5,A8,A8
.L2 B5,B8,B8
.L1 A5,A8,A8
.L2 B5,B8,B8
.L1 A5,A8,A8
.L2 B5,B8,B8
.L1 A5,A8,A8
.L2 B5,B8,B8
;********* load ai & ai + 1 from memory
;********* load bi & bi + 1 from memory
;**** pi = a0 b0
;**** pi1 = a1
; sum0 += (ai bi)
; sum1 += (ai+1
;***** branch to loop
;****** decrement loop counter
; pi = a0 b0
; pi1 = a1 b1
; sum0 += (ai bi)
; sum1 += (ai+1
; pi = a0 b0
; pi1 = a1 b1
; sum0 += (ai bi)
; sum1 += (ai+1
; pi = a0 b0
; pi1 = a1 b1
; sum0 += (ai bi)
; sum1 += (ai+1
; pi = a0 b0
; pi1 = a1 b1
; sum0 += (ai bi)
; sum1 += (ai+1
; pi = a0 b0
; pi1 = a1 b1
; sum0 += (ai bi)
; sum1 += (ai+1
; sum0 += (ai bi)
; sum1 += (ai+1
; sum0 += (ai bi)
; sum1 += (ai+1
; sum0 += (ai bi)
; sum1 += (ai+1
; sum0 += (ai bi)
; sum1 += (ai+1
Optimizing Assembly Code via Linear Assembly
Software Pipelining
b1
bi+1)
ADDSPs
1
bi+1)
2
bi+1)
3
bi+1)
4
bi+1)
5
bi+1)
6
bi+1)
7
bi+1)
8
bi+1)
9
bi+1)
MPYSPs
1
2
3
4
5
6-49