--- trunk/xvidcore/src/bitstream/x86_asm/cbp_mmx.asm 2002/07/07 09:45:40 262 +++ trunk/xvidcore/src/bitstream/x86_asm/cbp_mmx.asm 2005/09/24 03:10:37 1642 @@ -1,124 +1,150 @@ -;/************************************************************************** +;/**************************************************************************** ; * -; * XVID MPEG-4 VIDEO CODEC -; * mmx cbp calc +; * XVID MPEG-4 VIDEO CODEC +; * - MMX CBP computation - ; * -; * This program is an implementation of a part of one or more MPEG-4 -; * Video tools as specified in ISO/IEC 14496-2 standard. Those intending -; * 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. +; * Copyright (C) 2005 Carlo Bramini +; * 2001-2003 Peter Ross +; * 2002-2003 Pascal Massimino ; * -; * This program is free software; you can redistribute it and/or modify -; * it under the terms of the GNU General Public License as published by -; * the Free Software Foundation; either version 2 of the License, or -; * (at your option) any later version. +; * This program is free software ; you can redistribute it and/or modify +; * it under the terms of the GNU General Public License as published by +; * the Free Software Foundation ; either version 2 of the License, or +; * (at your option) any later version. ; * -; * This program is distributed in the hope that it will be useful, -; * but WITHOUT ANY WARRANTY; without even the implied warranty of -; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -; * GNU General Public License for more details. +; * This program is distributed in the hope that it will be useful, +; * but WITHOUT ANY WARRANTY ; without even the implied warranty of +; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +; * GNU General Public License for more details. ; * -; * You should have received a copy of the GNU General Public License -; * along with this program; if not, write to the Free Software -; * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. +; * You should have received a copy of the GNU General Public License +; * along with this program ; if not, write to the Free Software +; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA ; * -; *************************************************************************/ - -;/************************************************************************** -; * -; * History: -; * -; * 14.06.2002 some further tweaks by -Skal- -; * 24.11.2001 inital version; (c)2001 peter ross +; * $Id: cbp_mmx.asm,v 1.14 2005-09-24 03:10:37 suxen_drol Exp $ ; * -; *************************************************************************/ +; ***************************************************************************/ +BITS 32 -bits 32 - -section .data +;============================================================================= +; Macros +;============================================================================= %macro cglobal 1 %ifdef PREFIX - global _%1 - %define %1 _%1 + %ifdef MARK_FUNCS + global _%1:function %1.endfunc-%1 + %define %1 _%1:function %1.endfunc-%1 + %else + global _%1 + %define %1 _%1 + %endif %else - global %1 + %ifdef MARK_FUNCS + global %1:function %1.endfunc-%1 + %else + global %1 + %endif %endif %endmacro -align 16 - -ignore_dc dw 0, -1, -1, -1, -1, -1, -1, -1 +;============================================================================= +; Local data +;============================================================================= + +%ifdef FORMAT_COFF +SECTION .rodata +%else +SECTION .rodata align=16 +%endif + +ALIGN 16 + +mult_mask: + db 0x10,0x20,0x04,0x08,0x01,0x02,0x00,0x00 +ignore_dc: + dw 0, -1, -1, -1 + +;============================================================================= +; Code +;============================================================================= -section .text +SECTION .text cglobal calc_cbp_mmx -;=========================================================================== -; +;----------------------------------------------------------------------------- ; uint32_t calc_cbp_mmx(const int16_t coeff[6][64]); -; -;=========================================================================== - -align 16 -calc_cbp_mmx: - push ebx - push esi - - mov esi, [esp + 8 + 4] ; coeff - xor eax, eax ; cbp = 0 - mov edx, (1 << 5) - - movq mm7, [ignore_dc] - -.loop - movq mm0, [esi] - movq mm1, [esi+8] - pand mm0, mm7 +;----------------------------------------------------------------------------- - por mm0, [esi+16] - por mm1, [esi+24] - - por mm0, [esi+32] - por mm1, [esi+40] - - por mm0, [esi+48] - por mm1, [esi+56] - - 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 - movq mm1, mm0 - psrlq mm1, 32 - lea esi, [esi + 128] - - por mm0, mm1 - movd ebx, mm0 +%macro MAKE_LOAD 1 + por mm0, [eax-128*1+%1*8] + por mm1, [eax+128*0+%1*8] + por mm2, [eax+128*1+%1*8] + por mm3, [eax+128*2+%1*8] + por mm4, [eax+128*3+%1*8] + por mm5, [eax+128*4+%1*8] +%endmacro - test ebx, ebx - jz .next - or eax, edx ; cbp |= 1 << (5-i) +ALIGN 16 +calc_cbp_mmx: + mov eax, [esp + 4] ; coeff -.next - shr edx,1 - jnc .loop + movq mm7, [ignore_dc] + pxor mm6, mm6 ; used only for comparing + movq mm0, [eax+128*0] + movq mm1, [eax+128*1] + movq mm2, [eax+128*2] + movq mm3, [eax+128*3] + movq mm4, [eax+128*4] + movq mm5, [eax+128*5] + add eax, 8+128 + pand mm0, mm7 + pand mm1, mm7 + pand mm2, mm7 + pand mm3, mm7 + pand mm4, mm7 + pand mm5, mm7 + + MAKE_LOAD 0 + MAKE_LOAD 1 + MAKE_LOAD 2 + MAKE_LOAD 3 + MAKE_LOAD 4 + MAKE_LOAD 5 + MAKE_LOAD 6 + MAKE_LOAD 7 + MAKE_LOAD 8 + MAKE_LOAD 9 + MAKE_LOAD 10 + MAKE_LOAD 11 + MAKE_LOAD 12 + MAKE_LOAD 13 + MAKE_LOAD 14 + + movq mm7, [mult_mask] + packssdw mm0, mm1 + packssdw mm2, mm3 + packssdw mm4, mm5 + packssdw mm0, mm2 + packssdw mm4, mm6 + pcmpeqw mm0, mm6 + pcmpeqw mm4, mm6 + pcmpeqw mm0, mm6 + pcmpeqw mm4, mm6 + psrlw mm0, 15 + psrlw mm4, 15 + packuswb mm0, mm4 + pmaddwd mm0, mm7 + + movq mm1, mm0 + psrlq mm1, 32 + paddusb mm0, mm1 + + movd eax, mm0 + shr eax, 8 + and eax, 0x3F + ret +.endfunc - pop esi - pop ebx - - ret \ No newline at end of file