Motorola DSP96002 User Manual page 559

rep
mac
macr
move
_ecols
move
move
_erows
At a DSP56000/1 clock speed of 20.5 MHz, a [10x10][10x10] can be computed in .1715 ms.
move
move
move
move
do
move
move
do
move
fclr
fclr
rep
fmpy
move
_cols
move
_rows
At a DSP96002 clock speed of 26.66 MHz, a [10x10][10x10] can be computed in .1325 ms.
B-40
#N-1
x0,y0,a x:(r1)+,x0 y:(r5)+n5,y0
x0,y0,a (r4)+
a,y:(r6)+
(r0)+n0
#mat_b,r4
DSP96002 IMPLEMENTATION
#mat_a,r0
#mat_c,r6
#N,n0
n0,n5
#N,_rows
#mat_b,r4
r0,r1
#N,_cols
d0
d1
#N
d4,d5,d1 fadd.s d1,d0 x:(r1)+,d4.s y:(r5)+n5,d5.s 1
fadd.s d1,d0 r0,r1
DSP96002 USER'S MANUAL
;sum
;finish, next column B
;save output
;next row A
;first element B
((8+(N-1))N+5)N+8 = N 3 +7 * N 2 +5N+8
Program ICycles
Words
;point to A
;output mat C
;array size
;point to B
;copy start of row
r4,r5
(r4)+
x:(r1)+,d4.s y:(r5)+n5,d5.s 1
d0.s,y:(r6)+
(r0)+n0
Totals:
((N+7)N+6)N+7 = N 3 +7 * N 2 +6N+7
1 1
1 2
1 1
1 1
1 1
1 1
----- -----
19
1 1
1 1
1 1
1 1
2 3
1 1
1 1
2 3
1 1
1 1
1
1 2
1
1 1
1 1
1 1
----- -----
19
MOTOROLA