📄 fft16_mac.s.list
字号:
; float tr,ti;
;
; j=nd2;
; for (i=1;i<=n-2;i++){
; if (i<j){
; tr=ReX[j];
; ti=ImX[j];
; ReX[j]=ReX[i];
; ImX[j]=ImX[i];
; ReX[i]=tr;
; ImX[i]=ti;
; }
; k=nd2;
; while (k<=j){
; j=j-k;
; k=k>>1;
; }
; j=j+k;
; }
;
; }
_rev_addr_sort
0x00000000: 4cef03000004 movem.l (4,a7),a0/a1 ;load start addresses of ReX and ImX buffers
;to a0 and a1 registers
0x00000006: 243c00000200 move.l #512,d2 ;store nd2*2 into d2 register (d2 is j)
;'2' is the size of one sample (2 bytes)
0x0000000c: 7202 moveq.l #2,d1 ;d1 is i
for ;for (i=1;i<=n-2;i++){
0x0000000e: b282 cmp.l d2,d1 ;if (i<j){
0x00000010: 641c bcc.b jk
0x00000012: 45f01800 lea (a0,d1.l),a2 ;add index j to the beginning address of buffer ReX
;and store it into a2
0x00000016: 47f11800 lea (a1,d1.l),a3 ;add index j to the beginning address of buffer ImX
;and store it into a3
0x0000001a: 36302800 move.w (a0,d2.l),d3 ;tr=ReX[j];
0x0000001e: 38312800 move.w (a1,d2.l),d4 ;ti=ImX[j];
0x00000022: 31922800 move.w (a2),(a0,d2.l) ;ReX[j]=ReX[i];
0x00000026: 33932800 move.w (a3),(a1,d2.l) ;ImX[j]=ImX[i];
0x0000002a: 3483 move.w d3,(a2) ;ReX[i]=tr;
0x0000002c: 3684 move.w d4,(a3) ;ImX[i]=ti;
jk
0x0000002e: 203c00000200 move.l #512,d0 ;d0 is k, k=nd2*2
;'2' is the size of one sample (2 bytes);
0x00000034: b480 cmp.l d0,d2 ;while (k<=j){
0x00000036: 6508 bcs.b j
0x00000038: 9480 while sub.l d0,d2 ;j=j-k;
0x0000003a: e288 lsr.l #1,d0 ;k=k>>1;
0x0000003c: b480 cmp.l d0,d2
0x0000003e: 64f8 bcc.b while
j
0x00000040: d480 add.l d0,d2 ;j=j+k;
0x00000042: 5481 addq.l #2,d1
0x00000044: 0c81000003fe cmp.l #1022,d1
0x0000004a: 65c2 bcs.b for
0x0000004c: 4e75 rts
;**********************************************************************************
;Real FFT
;**********************************************************************************
;Upon entry, REX[ ] contains the real input signal, while values in IMX[ ] are
;ignored.
;Upon return, REX[ ] and IMX[ ] contain the DFT output.
;All signals run from 0 to 1023.
;**********************************************************************************
_fft16_mac
0x0000004e: 4fefffb8 lea -72(a7),a7
0x00000052: 48d77fff movem.l d0-d7/a0-a6,(a7) ;store contents of all registers into stack
;Corresponding C code: ;separate even and odd points
;for (i=0;i<512;i++){
; ReX[i]=ReX[2*i];
; ImX[i]=ReX[2*i+1];
;}
0x00000056: 4cef0300004c movem.l (76,a7),a0/a1 ;separate even and odd points
;point a0 and a1 to ReX and ImX buffers
0x0000005c: 7000 moveq.l #0,d0
0x0000005e: 2448 movea.l a0,a2
reorder
0x00000060: 22300800 move.l (0,a0,d0.l),d1 ;ReX[i],ImX[i] - > d1
0x00000064: 32c1 move.w d1,(a1)+ ;ImX[i]=ImX[2*i+1];
0x00000066: 4841 swap d1
0x00000068: 34c1 move.w d1,(a2)+ ;ReX[i]=ReX[2*i];
0x0000006a: 5880 addq.l #4,d0 ;modification of loop counter
0x0000006c: 0c8000000800 cmpi.l #2048,d0
0x00000072: 65ec bcs.b reorder
0x00000074: 43e9fc00 lea -1024(a1),a1 ;bit reversal sorting
0x00000078: 2f09 move.l a1,-(a7) ;push address of ImX[] buffer into the stack
0x0000007a: 2f08 move.l a0,-(a7) ;push address of ReX[] buffer into the stack
0x0000007c: 4eb900000000 jsr _rev_addr_sort ;jump to subroutine
0x00000082: 4fef0008 lea (8,a7),a7 ;
0x00000086: 7c00 moveq.l #0,d6 ;first stage of FFT
first_stage
0x00000088: 2010 move.l (a0),d0 ;d0 = ar,br
0x0000008a: 2400 move.l d0,d2 ;d2 = ar,br
0x0000008c: 4840 swap d0 ;d0 = br,ar
0x0000008e: 48c0 ext.l d0 ;d0 = ar
0x00000090: 48c2 ext.l d2 ;d2 = br
0x00000092: 2211 move.l (a1),d1 ;d1 = ai,bi
0x00000094: 2601 move.l d1,d3 ;d3 = ai,bi
0x00000096: 4841 swap d1 ;d1 = bi,ai
0x00000098: 48c1 ext.l d1 ;d1 = ai
0x0000009a: 48c3 ext.l d3 ;d3 = bi
0x0000009c: 2802 move.l d2,d4 ;d4 = br
0x0000009e: 2a03 move.l d3,d5 ;d5 = bi
;on the first stage the butterfly operation
;looks like:
;xr = ar + br
;xi = ai + bi
;yr = ar - br
;yi = ai - bi
0x000000a0: d480 add.l d0,d2 ;xr = ar + br
0x000000a2: 30c2 move.w d2,(a0)+
0x000000a4: d681 add.l d1,d3 ;xi = ai + bi
0x000000a6: 32c3 move.w d3,(a1)+
0x000000a8: 9084 sub.l d4,d0 ;yr = ar - br
0x000000aa: 30c0 move.w d0,(a0)+
0x000000ac: 9285 sub.l d5,d1 ;yi = ai - bi
0x000000ae: 32c1 move.w d1,(a1)+
0x000000b0: 5286 addq.l #1,d6
0x000000b2: 0c8600000100 cmpi.l #256,d6
0x000000b8: 65ce bcs.b first_stage
0x000000ba: 2c7c00000000 movea.l #0,a6 ;second stage of FFT
0x000000c0: 41e8fc00 lea (-1024,a0),a0 ;a0 points to the beginning of ReX buffer
0x000000c4: 43e9fc00 lea (-1024,a1),a1 ;a1 points to the beginning of ImX buffer
;on the second stage we will calculate
;two butterflies, first of which looks like
;butterfly on the first stage
;xr0 = ar0 + br0
;xi0 = ai0 + bi0
;yr0 = ar0 - br0
;yi0 = ai0 - bi0,
;and second looks like
;xr1 = ar1 + bi1
;xi1 = ai1 - br1
;yr1 = ar1 - bi1
;yi1 = ai1 + br1
second_stage
0x000000c8: 2010 move.l (a0),d0 ;d0 = ar0,ar1
0x000000ca: 2200 move.l d0,d1 ;d1 = ar0,ar1
0x000000cc: 4840 swap d0 ;d0 = ar1,ar0
0x000000ce: 48c0 ext.l d0 ;d0 = ar0
0x000000d0: 48c1 ext.l d1 ;d1 = ar1
0x000000d2: 24280004 move.l (4,a0),d2 ;d2 = br0,br1
0x000000d6: 2602 move.l d2,d3 ;d3 = br0,br1
0x000000d8: 4842 swap d2 ;d2 = br1,br0
0x000000da: 48c2 ext.l d2 ;d2 = br0
0x000000dc: 48c3 ext.l d3 ;d3 = br1
0x000000de: 2811 move.l (a1),d4 ;d4 = ai0,ai1
0x000000e0: 2a04 move.l d4,d5 ;d5 = ai0,ai1
0x000000e2: 4844 swap d4 ;d4 = ai1,ai0
0x000000e4: 48c4 ext.l d4 ;d4 = ai0
0x000000e6: 48c5 ext.l d5 ;d5 = ai1
0x000000e8: 2c290004 move.l (4,a1),d6 ;d6 = bi0,bi1
0x000000ec: 2e06 move.l d6,d7 ;d7 = bi0,bi1
0x000000ee: 4846 swap d6 ;d6 = bi1,bi0
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -