Final Assembly With Move Instructions - Texas Instruments TMS320C6000 Programmer's Manual

Live-Too-Long Issues

6.10.6 Final Assembly With Move Instructions

Example 6–59. Assembly Code for Live-Too-Long With Move Instructions
LDH .D1
|| LDH .D2
MVK .S2
LDH .D1
|| LDH .D2
ZERO .S1
|| ZERO .S2
LDH .D1
|| LDH .D2
[B2] SUB .S2
MPY .M1
|| MPY .M2X
|| LDH .D1
|| LDH .D2
[B2] SUB .S2
||[B2] B .S1
SHR .S1
|| SHR .S2
|| MPY .M1
|| MPY .M2X
|| LDH .D1
|| LDH .D2
MPY .M1X
|| MV .D1
|| MPY .M2X
|| MV .D2
||[B2] SUB .S2
||[B2] B .S1
SHR .S1
|| SHR .S2
|| MPY .M1
|| MPY .M2X
|| LDH .D1
|| LDH .D2
6-108
Example 6–59 shows the final assembly code after software pipelining. The
performance of this loop is 212 cycles (2
*A4++,A0 ; load ai from memory
*B4++,B0 ; load bi from memory
100,B2 ; set up loop counter
*A4++,A0 ;* load ai from memory
*B4++,B0 ;* load bi from memory
A1 ; zero out accumulator
B1 ; zero out accumulator
*A4++,A0 ;** load ai from memory
*B4++,B0 ;** load bi from memory
B2,1,B2 ; decrement loop counter
A0,A6,A3 ; a0 = ai * c
B0,A6,B10 ; b0 = bi * c
*A4++,A0 ;*** load ai from memory
*B4++,B0 ;*** load bi from memory
B2,1,B2 ; decrement loop counter
LOOP ; branch to loop
A3,15,A5 ; a1 = a0 >> 15
B10,15,B5 ; b1 = b0 >> 15
A0,A6,A3 ;* a0 = ai * c
B0,A6,B10 ;* b0 = bi * c
*A4++,A0 ;**** load ai from memory
*B4++,B0 ;**** load bi from memory
A5,B6,A7 ; a2 = a1 * d
A3,A2 ; save a0 across iterations
B5,A8,B7 ; b2 = b1 * e
B10,B8 ; save b0 across iterations
B2,1,B2 ;* decrement loop counter
LOOP ;* branch to loop
A3,15,A5 ;* a1 = a0 >> 15
B10,15,B5 ;* b1 = b0 >> 15
A0,A6,A3 ;** a0 = ai * c
B0,A6,B10 ;** b0 = bi * c
*A4++,A0 ;***** load ai from memory
*B4++,B0 ;***** load bi from memory
100 + 11 + 1).