| 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) 2005 Carlo Bramini <carlo.bramix@libero.it> |
| 7 |
; * Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
; * 2001-2003 Peter Ross <pross@xvid.org> |
| 8 |
; * to use this software module in hardware or software products are |
; * 2002-2003 Pascal Massimino <skal@planet-d.net> |
|
; * 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. |
|
| 9 |
; * |
; * |
| 10 |
; * This program is free software; you can redistribute it and/or modify |
; * This program is free software; you can redistribute it and/or modify |
| 11 |
; * 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 |
| 19 |
; * |
; * |
| 20 |
; * You should have received a copy of the GNU General Public License |
; * You should have received a copy of the GNU General Public License |
| 21 |
; * along with this program; if not, write to the Free Software |
; * along with this program; if not, write to the Free Software |
| 22 |
; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
| 23 |
; * |
; * |
| 24 |
; *************************************************************************/ |
; * $Id: cbp_mmx.asm,v 1.18 2008-12-04 14:41:50 Isibaar Exp $ |
|
|
|
|
;/************************************************************************** |
|
|
; * |
|
|
; * History: |
|
| 25 |
; * |
; * |
| 26 |
; * 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> |
|
|
; * |
|
|
; *************************************************************************/ |
|
| 27 |
|
|
| 28 |
|
;============================================================================= |
| 29 |
|
; Macros |
| 30 |
|
;============================================================================= |
| 31 |
|
|
| 32 |
bits 32 |
%include "nasm.inc" |
| 33 |
|
|
| 34 |
section .data |
;============================================================================= |
| 35 |
|
; Local data |
| 36 |
|
;============================================================================= |
| 37 |
|
|
| 38 |
%macro cglobal 1 |
DATA |
|
%ifdef PREFIX |
|
|
global _%1 |
|
|
%define %1 _%1 |
|
|
%else |
|
|
global %1 |
|
|
%endif |
|
|
%endmacro |
|
| 39 |
|
|
| 40 |
align 16 |
ALIGN SECTION_ALIGN |
| 41 |
|
|
| 42 |
ignore_dc dw 0, -1, -1, -1, -1, -1, -1, -1 |
mult_mask: |
| 43 |
|
db 0x10,0x20,0x04,0x08,0x01,0x02,0x00,0x00 |
| 44 |
|
ignore_dc: |
| 45 |
|
dw 0, -1, -1, -1 |
| 46 |
|
|
| 47 |
|
;============================================================================= |
| 48 |
|
; Code |
| 49 |
|
;============================================================================= |
| 50 |
|
|
| 51 |
section .text |
TEXT |
| 52 |
|
|
| 53 |
|
cglobal calc_cbp_mmx |
| 54 |
|
|
| 55 |
;=========================================================================== |
;----------------------------------------------------------------------------- |
|
; |
|
| 56 |
; uint32_t calc_cbp_mmx(const int16_t coeff[6][64]); |
; uint32_t calc_cbp_mmx(const int16_t coeff[6][64]); |
| 57 |
; |
;----------------------------------------------------------------------------- |
|
;=========================================================================== |
|
| 58 |
|
|
| 59 |
align 16 |
%macro MAKE_LOAD 2 |
| 60 |
cglobal calc_cbp_mmx |
por mm0, [%2-128*1+%1*8] |
| 61 |
calc_cbp_mmx |
por mm1, [%2+128*0+%1*8] |
| 62 |
push ebx |
por mm2, [%2+128*1+%1*8] |
| 63 |
push ecx |
por mm3, [%2+128*2+%1*8] |
| 64 |
push edx |
por mm4, [%2+128*3+%1*8] |
| 65 |
push esi |
por mm5, [%2+128*4+%1*8] |
| 66 |
|
%endmacro |
| 67 |
|
|
| 68 |
mov esi, [esp + 16 + 4] ; coeff |
ALIGN SECTION_ALIGN |
| 69 |
movq mm7, [ignore_dc] |
calc_cbp_mmx: |
| 70 |
|
mov _EAX, prm1 ; coeff |
| 71 |
|
|
| 72 |
xor eax, eax ; cbp = 0 |
movq mm7, [ignore_dc] |
| 73 |
mov edx, 6 |
pxor mm6, mm6 ; used only for comparing |
| 74 |
.loop |
movq mm0, [_EAX+128*0] |
| 75 |
movq mm0, [esi] |
movq mm1, [_EAX+128*1] |
| 76 |
|
movq mm2, [_EAX+128*2] |
| 77 |
|
movq mm3, [_EAX+128*3] |
| 78 |
|
movq mm4, [_EAX+128*4] |
| 79 |
|
movq mm5, [_EAX+128*5] |
| 80 |
|
add _EAX, 8+128 |
| 81 |
pand mm0, mm7 |
pand mm0, mm7 |
| 82 |
movq mm1, [esi+8] |
pand mm1, mm7 |
| 83 |
|
pand mm2, mm7 |
| 84 |
por mm0, [esi+16] |
pand mm3, mm7 |
| 85 |
por mm1, [esi+24] |
pand mm4, mm7 |
| 86 |
|
pand mm5, mm7 |
| 87 |
por mm0, [esi+32] |
|
| 88 |
por mm1, [esi+40] |
MAKE_LOAD 0, _EAX |
| 89 |
|
MAKE_LOAD 1, _EAX |
| 90 |
por mm0, [esi+48] |
MAKE_LOAD 2, _EAX |
| 91 |
por mm1, [esi+56] |
MAKE_LOAD 3, _EAX |
| 92 |
|
MAKE_LOAD 4, _EAX |
| 93 |
|
MAKE_LOAD 5, _EAX |
| 94 |
|
MAKE_LOAD 6, _EAX |
| 95 |
|
MAKE_LOAD 7, _EAX |
| 96 |
|
MAKE_LOAD 8, _EAX |
| 97 |
|
MAKE_LOAD 9, _EAX |
| 98 |
|
MAKE_LOAD 10, _EAX |
| 99 |
|
MAKE_LOAD 11, _EAX |
| 100 |
|
MAKE_LOAD 12, _EAX |
| 101 |
|
MAKE_LOAD 13, _EAX |
| 102 |
|
MAKE_LOAD 14, _EAX |
| 103 |
|
|
| 104 |
|
movq mm7, [mult_mask] |
| 105 |
|
packssdw mm0, mm1 |
| 106 |
|
packssdw mm2, mm3 |
| 107 |
|
packssdw mm4, mm5 |
| 108 |
|
packssdw mm0, mm2 |
| 109 |
|
packssdw mm4, mm6 |
| 110 |
|
pcmpeqw mm0, mm6 |
| 111 |
|
pcmpeqw mm4, mm6 |
| 112 |
|
pcmpeqw mm0, mm6 |
| 113 |
|
pcmpeqw mm4, mm6 |
| 114 |
|
psrlw mm0, 15 |
| 115 |
|
psrlw mm4, 15 |
| 116 |
|
packuswb mm0, mm4 |
| 117 |
|
pmaddwd mm0, mm7 |
| 118 |
|
|
|
por mm0, [esi+64] |
|
|
por mm1, [esi+72] |
|
|
|
|
|
por mm0, [esi+80] |
|
|
por mm1, [esi+88] |
|
|
|
|
|
por mm0, [esi+96] |
|
|
por mm1, [esi+104] |
|
|
|
|
|
por mm0, [esi+112] |
|
|
por mm1, [esi+120] |
|
|
|
|
|
por mm0, mm1 |
|
| 119 |
movq mm1, mm0 |
movq mm1, mm0 |
| 120 |
psrlq mm1, 32 |
psrlq mm1, 32 |
| 121 |
por mm0, mm1 |
paddusb mm0, mm1 |
|
movd ebx, mm0 |
|
|
|
|
|
add esi, 128 |
|
|
|
|
|
or ebx, ebx |
|
|
jz .iterate |
|
|
|
|
|
; cbp |= 1 << (edx-1) |
|
|
|
|
|
; 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 |
|
|
|
|
|
pop esi |
|
|
pop edx |
|
|
pop ecx |
|
|
pop ebx |
|
| 122 |
|
|
| 123 |
|
movd eax, mm0 |
| 124 |
|
shr _EAX, 8 |
| 125 |
|
and _EAX, 0x3F |
| 126 |
ret |
ret |
| 127 |
|
ENDFUNC |
| 128 |
|
|
| 129 |
|
|
| 130 |
|
%ifidn __OUTPUT_FORMAT__,elf |
| 131 |
|
section ".note.GNU-stack" noalloc noexec nowrite progbits |
| 132 |
|
%endif |
| 133 |
|
|
|
;=========================================================================== |
|
|
; |
|
|
; 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 |
|
Parent Directory
| 
Revision Log
| 
Patch