1 |
;/************************************************************************** |
;/**************************************************************************** |
2 |
; * |
; * |
3 |
; * XVID MPEG-4 VIDEO CODEC |
; * XVID MPEG-4 VIDEO CODEC |
4 |
; * mmx cbp calc |
; * - MMX CBP computation - |
5 |
; * |
; * |
6 |
; * This program is an implementation of a part of one or more MPEG-4 |
; * Copyright (C) 2001-2003 Peter Ross <pross@xvid.org> |
7 |
; * Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
; * 2002-2003 Pascal Massimino <skal@planet-d.net> |
|
; * to use this software module in hardware or software products are |
|
|
; * advised that its use may infringe existing patents or copyrights, and |
|
|
; * any such use would be at such party's own risk. The original |
|
|
; * developer of this software module and his/her company, and subsequent |
|
|
; * editors and their companies, will have no liability for use of this |
|
|
; * software or modifications or derivatives thereof. |
|
8 |
; * |
; * |
9 |
; * This program is free software; you can redistribute it and/or modify |
; * This program is free software; you can redistribute it and/or modify |
10 |
; * it under the terms of the GNU General Public License as published by |
; * it under the terms of the GNU General Public License as published by |
18 |
; * |
; * |
19 |
; * You should have received a copy of the GNU General Public License |
; * You should have received a copy of the GNU General Public License |
20 |
; * along with this program; if not, write to the Free Software |
; * along with this program; if not, write to the Free Software |
21 |
; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
22 |
; * |
; * |
23 |
; *************************************************************************/ |
; * $Id$ |
|
|
|
|
;/************************************************************************** |
|
|
; * |
|
|
; * History: |
|
24 |
; * |
; * |
25 |
; * 24.04.2002 had to use sse2's movdqu instead of movdqa (???) |
; ***************************************************************************/ |
|
; * 17.04.2002 sse2 stuff |
|
|
; * 22.03.2002 0.01 ; Min Chen <chenm001@163.com> |
|
|
; * ; use 386 cpu's 'BTS' to replace 'cbp |= 1 << (edx-1)' |
|
|
; * 24.11.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
|
|
; * |
|
|
; *************************************************************************/ |
|
|
|
|
26 |
|
|
27 |
bits 32 |
BITS 32 |
28 |
|
|
29 |
section .data |
;============================================================================= |
30 |
|
; Macros |
31 |
|
;============================================================================= |
32 |
|
|
33 |
%macro cglobal 1 |
%macro cglobal 1 |
34 |
%ifdef PREFIX |
%ifdef PREFIX |
39 |
%endif |
%endif |
40 |
%endmacro |
%endmacro |
41 |
|
|
42 |
align 16 |
;============================================================================= |
43 |
|
; Local data |
44 |
|
;============================================================================= |
45 |
|
|
46 |
ignore_dc dw 0, -1, -1, -1, -1, -1, -1, -1 |
SECTION .rodata |
47 |
|
|
48 |
|
ALIGN 16 |
49 |
|
|
50 |
section .text |
ignore_dc: |
51 |
|
dw 0, -1, -1, -1, -1, -1, -1, -1 |
52 |
|
|
53 |
|
;============================================================================= |
54 |
|
; Code |
55 |
|
;============================================================================= |
56 |
|
|
57 |
;=========================================================================== |
SECTION .text |
|
; |
|
|
; uint32_t calc_cbp_mmx(const int16_t coeff[6][64]); |
|
|
; |
|
|
;=========================================================================== |
|
58 |
|
|
|
align 16 |
|
59 |
cglobal calc_cbp_mmx |
cglobal calc_cbp_mmx |
60 |
calc_cbp_mmx |
|
61 |
|
;----------------------------------------------------------------------------- |
62 |
|
; uint32_t calc_cbp_mmx(const int16_t coeff[6][64]); |
63 |
|
;----------------------------------------------------------------------------- |
64 |
|
|
65 |
|
ALIGN 16 |
66 |
|
calc_cbp_mmx: |
67 |
push ebx |
push ebx |
|
push ecx |
|
|
push edx |
|
68 |
push esi |
push esi |
69 |
|
|
70 |
mov esi, [esp + 16 + 4] ; coeff |
mov esi, [esp + 8 + 4] ; coeff |
71 |
|
xor eax, eax ; cbp = 0 |
72 |
|
mov edx, (1 << 5) |
73 |
|
|
74 |
movq mm7, [ignore_dc] |
movq mm7, [ignore_dc] |
75 |
|
|
|
xor eax, eax ; cbp = 0 |
|
|
mov edx, 6 |
|
76 |
.loop |
.loop |
77 |
movq mm0, [esi] |
movq mm0, [esi] |
|
pand mm0, mm7 |
|
78 |
movq mm1, [esi+8] |
movq mm1, [esi+8] |
79 |
|
pand mm0, mm7 |
80 |
|
|
81 |
por mm0, [esi+16] |
por mm0, [esi+16] |
82 |
por mm1, [esi+24] |
por mm1, [esi+24] |
102 |
por mm0, mm1 |
por mm0, mm1 |
103 |
movq mm1, mm0 |
movq mm1, mm0 |
104 |
psrlq mm1, 32 |
psrlq mm1, 32 |
105 |
|
lea esi, [esi + 128] |
106 |
|
|
107 |
por mm0, mm1 |
por mm0, mm1 |
108 |
movd ebx, mm0 |
movd ebx, mm0 |
109 |
|
|
110 |
add esi, 128 |
test ebx, ebx |
111 |
|
jz .next |
112 |
or ebx, ebx |
or eax, edx ; cbp |= 1 << (5-i) |
113 |
jz .iterate |
|
114 |
|
.next |
115 |
; cbp |= 1 << (edx-1) |
shr edx,1 |
116 |
|
jnc .loop |
|
; Change by Chenm001 <chenm001@163.com> |
|
|
;mov ecx, edx |
|
|
;dec ecx |
|
|
;mov ebx, 1 |
|
|
;shl ebx, cl |
|
|
;or eax, ebx |
|
|
lea ebx,[edx-1] |
|
|
bts eax,ebx |
|
|
|
|
|
.iterate dec edx |
|
|
jnz .loop |
|
117 |
|
|
118 |
pop esi |
pop esi |
|
pop edx |
|
|
pop ecx |
|
119 |
pop ebx |
pop ebx |
120 |
|
|
121 |
ret |
ret |
|
|
|
|
|
|
|
|
|
|
;=========================================================================== |
|
|
; |
|
|
; uint32_t calc_cbp_sse2(const int16_t coeff[6][64]); |
|
|
; |
|
|
; not enabled - slower than mmx? |
|
|
; |
|
|
;=========================================================================== |
|
|
|
|
|
align 16 |
|
|
cglobal calc_cbp_sse2 |
|
|
calc_cbp_sse2 |
|
|
push esi |
|
|
|
|
|
mov esi, [esp + 4 + 4] ; coeff |
|
|
movdqu xmm7, [ignore_dc] ; mask to ignore dc value |
|
|
|
|
|
xor eax, eax ; cbp = 0 |
|
|
pxor xmm6, xmm6 ; zeroes to help psadbw |
|
|
|
|
|
.first movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 ; xmm0 = xmm1 = 128 bits worth of info |
|
|
psadbw xmm0, xmm6 ; contains 2 dwords with sums |
|
|
movhlps xmm1, xmm0 ; move high dword from xmm0 to low xmm1 |
|
|
por xmm0, xmm1 ; combine |
|
|
movd ecx, xmm0 ; if ecx set, values were found |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .second |
|
|
|
|
|
bts eax, 5 |
|
|
|
|
|
.second movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 |
|
|
psadbw xmm0, xmm6 |
|
|
movhlps xmm1, xmm0 |
|
|
por xmm0, xmm1 |
|
|
movd ecx, xmm0 |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .third |
|
|
|
|
|
bts eax, 4 |
|
|
|
|
|
.third movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 |
|
|
psadbw xmm0, xmm6 |
|
|
movhlps xmm1, xmm0 |
|
|
por xmm0, xmm1 |
|
|
movd ecx, xmm0 |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .fourth |
|
|
|
|
|
bts eax, 3 |
|
|
|
|
|
.fourth movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 |
|
|
psadbw xmm0, xmm6 |
|
|
movhlps xmm1, xmm0 |
|
|
por xmm0, xmm1 |
|
|
movd ecx, xmm0 |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .fifth |
|
|
|
|
|
bts eax, 2 |
|
|
|
|
|
.fifth movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 |
|
|
psadbw xmm0, xmm6 |
|
|
movhlps xmm1, xmm0 |
|
|
por xmm0, xmm1 |
|
|
movd ecx, xmm0 |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .sixth |
|
|
|
|
|
bts eax, 1 |
|
|
|
|
|
.sixth movdqa xmm0, [esi] |
|
|
pand xmm0, xmm7 |
|
|
movdqa xmm1, [esi+16] |
|
|
|
|
|
por xmm0, [esi+32] |
|
|
por xmm1, [esi+48] |
|
|
por xmm0, [esi+64] |
|
|
por xmm1, [esi+80] |
|
|
por xmm0, [esi+96] |
|
|
por xmm1, [esi+112] |
|
|
|
|
|
por xmm0, xmm1 |
|
|
psadbw xmm0, xmm6 |
|
|
movhlps xmm1, xmm0 |
|
|
por xmm0, xmm1 |
|
|
movd ecx, xmm0 |
|
|
|
|
|
or ecx, ecx |
|
|
jz .end |
|
|
|
|
|
bts eax, 0 |
|
|
|
|
|
.end pop esi |
|
|
|
|
|
ret |
|