InterCD 2001 May

home *** CD-ROM | disk | FTP | other *** search

/ InterCD 2001 May / may_2001.iso / intercd / root / Multimedia / ^DivX_Article / virtualdub / VirtualDub-source-1_4d / a_sharpen.asm < prev next >

Wrap

Assembly Source File | 2001-03-20 | 8.9 KB | 587 lines

; VirtualDub - Video processing and capture application ; Copyright (C) 1998-2001 Avery Lee ; ; 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. ; ; 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. ; ; Why did I write yet another ASM module? ; ; Because I took the OBJ output of Visual C++ 4.0 on maximum optimization, ; disassembled it, and said out loud, "God this is sh*tty code!" ; (Without the asterisk, to boot.) ; That is why. ; ; Hand assembly optimization forever. ; .586 .mmx .model flat .code extern _MMX_enabled : byte public _asm_sharpen_run ;asm_sharpen_run( ; [esp+ 4] void *dst, ; [esp+ 8] void *src, ; [esp+12] ulong width, ; [esp+16] ulong height, ; [esp+20] ulong srcstride, ; [esp+24] ulong dststride, ; [esp+28] long a_mult, ; [esp+32] long b_mult); _asm_sharpen_run: test _MMX_enabled,1 jnz _asm_sharpen_run_MMX push ebp push edi push esi push edx push ecx push ebx push eax mov esi,[esp+ 8+28] mov edi,[esp+20+28] mov edx,[esp+ 4+28] mov ebp,[esp+16+28] rowloop: push ebp mov ebp,[esp+12+32] mov eax,ebp shl eax,2 add esi,eax colloop: push ebp push edx mov eax,[esi-4] mov ebx,[esi+4-4] mov ecx,eax mov edx,ebx and eax,00ff00ffh and ebx,00ff00ffh and ecx,0000ff00h and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi+edi+4-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi+edi*2-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi+edi*2+4-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi-4-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi+edi-4-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx mov ebx,[esi+edi*2-4-4] mov edx,ebx and ebx,00ff00ffh and edx,0000ff00h add eax,ebx add ecx,edx ;EAX now holds the neighbor sum of red and blue ;ECX now holds the neighbor sum of green mov ebx,eax mov ebp,[esi+edi-4] shr ebx,16 and eax,0000ffffh imul eax,[esp+28+40] imul ebx,[esp+28+40] imul ecx,[esp+28+40] push esi push edi mov edx,ebp mov esi,ebp shr ebp,16 and edx,000000ffh and ebp,000000ffh and esi,0000ff00h imul edx,[esp+32+48] imul esi,[esp+32+48] imul ebp,[esp+32+48] add eax,edx add ecx,esi add ebx,ebp mov edx,edx ;(ESI,EDX): ; 00010000-7FFFFFFF 11 ; 00000000-0000FFFF 10 ; 80000000-FFFFFFFF 00 mov esi,eax ;b1 u mov edi,ebx ;r1 v xor esi,-1 ;b2 u xor edi,-1 ;r2 v mov edx,esi ;b3 u mov ebp,edi ;r3 v sar esi,31 ;b4 u add edx,00010000h ;b5 v sar edi,31 ;r4 u add ebp,00010000h ;r5 v sar edx,31 ;b6 u and eax,esi ;b7 v sar ebp,31 ;r6 u mov esi,ecx ;g1 v and ebx,edi ;r7 u xor esi,-1 ;g2 v or eax,edx ;b8 u mov edx,esi ;g3 v sar esi,31 ;g4 u add edx,01000000h ;g5 v and ecx,esi ;g6 u or ebx,ebp ;r8 v sar edx,31 ;g7 u and ebx,0000ff00h ; v shl ebx,8 ; u or ecx,edx ;g8 v shr eax,8 ; u and ecx,00ff0000h ; v shr ecx,8 ; u pop edi ; v pop esi ; u and eax,000000ffh ; v pop edx ; u or eax,ecx ; v pop ebp ; u or eax,ebx ; v sub esi,4 ; u mov [edx+ebp*4-4],eax ;v dec ebp jne colloop pop ebp add esi,edi add edx,[esp+24+28] dec ebp jne rowloop pop eax pop ebx pop ecx pop edx pop esi pop edi pop ebp ret ;********* _asm_sharpen_run_MMX: push ebp push edi push esi push edx push ecx push ebx push eax mov esi,[esp+ 8+28] mov edi,[esp+20+28] mov edx,[esp+ 4+28] mov ebp,[esp+16+28] ;mm5: neighbor multiplier ;mm6: center multiplier ;mm7: zero pxor mm7,mm7 mov eax,[esp+28+28] neg eax shr eax,2 mov ebx,eax shl ebx,16 or eax,ebx movd mm0,eax movd mm5,eax psllq mm5,32 por mm5,mm0 mov eax,[esp+32+28] shr eax,2 and eax,0000fff8h mov ebx,eax shl ebx,16 or eax,ebx movd mm0,eax movd mm6,eax psllq mm6,32 por mm6,mm0 rowloop_MMX: push ebp push edx push esi mov ebp,[esp+12+40] movd mm0,[esi-4] punpcklbw mm0,mm7 movq mm1,[esi] movq mm2,mm1 punpcklbw mm1,mm7 punpckhbw mm2,mm7 paddw mm0,mm1 paddw mm0,mm2 movd mm1,[esi+edi*2-4] punpcklbw mm1,mm7 paddw mm0,mm1 movq mm1,[esi+edi*2] movq mm2,mm1 punpcklbw mm1,mm7 punpckhbw mm2,mm7 paddw mm0,mm1 paddw mm0,mm2 movd mm1,[esi+edi-4] punpcklbw mm1,mm7 movd mm2,[esi+edi+4] punpcklbw mm2,mm7 paddw mm1,mm2 paddw mm1,mm0 pmullw mm1,mm5 movd mm2,[esi+edi] punpcklbw mm2,mm7 pmullw mm2,mm6 psubusw mm2,mm1 psrlw mm2,6 packuswb mm2,mm2 movd [edx],mm2 add esi,4 dec ebp movd mm1,[esi+4] add edx,4 movd mm4,[esi+edi*2+4] punpcklbw mm1,mm7 punpcklbw mm4,mm7 paddw mm0,mm1 movd mm1,[esi-8] movd mm2,[esi+edi*2-8] punpcklbw mm1,mm7 paddw mm0,mm4 jmp colloop_MMX_entry colloop_MMX: movd mm1,[esi+4] add edx,4 movd mm4,[esi+edi*2+4] punpcklbw mm1,mm7 psubusw mm3,mm2 punpcklbw mm4,mm7 psrlw mm3,6 paddw mm0,mm1 movd mm1,[esi-8] packuswb mm3,mm3 movd mm2,[esi+edi*2-8] punpcklbw mm1,mm7 movd [edx-4],mm3 paddw mm0,mm4 colloop_MMX_entry: punpcklbw mm2,mm7 psubw mm0,mm1 movd mm1,[esi+edi-4] psubw mm0,mm2 movd mm2,[esi+edi+4] punpcklbw mm1,mm7 movd mm3,[esi+edi] punpcklbw mm2,mm7 punpcklbw mm3,mm7 paddw mm1,mm0 pmullw mm3,mm6 paddw mm2,mm1 pmullw mm2,mm5 add esi,4 dec ebp jne colloop_MMX psubusw mm3,mm2 psrlw mm3,6 packuswb mm3,mm3 movd [edx],mm3 pop esi pop edx pop ebp add esi,edi add edx,[esp+24+28] dec ebp jne rowloop_MMX pop eax pop ebx pop ecx pop edx pop esi pop edi pop ebp emms ret IF 0 ;******************************** movd mm0,[esi-4] punpcklbw mm0,mm7 movq mm1,[esi] movq mm2,mm1 punpcklbw mm1,mm7 punpckhbw mm2,mm7 paddw mm0,mm1 paddw mm0,mm2 movd mm1,[esi+edi*2-4] punpcklbw mm1,mm7 paddw mm0,mm1 movq mm1,[esi+edi*2] movq mm2,mm1 punpcklbw mm1,mm7 punpckhbw mm2,mm7 paddw mm0,mm1 paddw mm0,mm2 movd mm1,[esi+edi-4] punpcklbw mm1,mm7 movd mm2,[esi+edi+4] punpcklbw mm2,mm7 paddw mm1,mm2 paddw mm1,mm0 pmullw mm1,mm5 movd mm2,[esi+edi] punpcklbw mm2,mm7 pmullw mm2,mm6 psubw mm2,mm1 psrlw mm2,6 packuswb mm2,mm2 movd [edx],mm2 add esi,4 dec ebp movd mm1,[esi+4] movd mm2,[esi+edi*2+4] punpcklbw mm1,mm7 punpcklbw mm2,mm7 paddw mm0,mm1 add edx,4 jmp colloop_MMX_entry colloop_MMX: movd mm1,[esi+4] psubusw mm3,mm2 movd mm2,[esi+edi*2+4] psrlw mm3,6 punpcklbw mm1,mm7 packuswb mm3,mm3 punpcklbw mm2,mm7 paddw mm0,mm1 movd [edx-4],mm3 add edx,4 colloop_MMX_entry: movd mm1,[esi-8] paddw mm0,mm2 movd mm2,[esi+edi*2-8] punpcklbw mm1,mm7 punpcklbw mm2,mm7 psubw mm0,mm1 movd mm1,[esi+edi-4] psubw mm0,mm2 movd mm2,[esi+edi+4] punpcklbw mm1,mm7 movd mm3,[esi+edi] punpcklbw mm2,mm7 punpcklbw mm3,mm7 paddw mm1,mm0 pmullw mm3,mm6 paddw mm2,mm1 pmullw mm2,mm5 add esi,4 dec ebp jne colloop_MMX ;***************************************** movd mm1,[esi+4] add edx,4 movd mm2,[esi+edi*2+4] punpcklbw mm1,mm7 punpcklbw mm2,mm7 paddw mm0,mm1 movd mm1,[esi-8] paddw mm0,mm2 movd mm2,[esi+edi*2-8] punpcklbw mm1,mm7 punpcklbw mm2,mm7 psubw mm0,mm1 movd mm1,[esi+edi-4] psubw mm0,mm2 movd mm2,[esi+edi+4] punpcklbw mm1,mm7 movd mm3,[esi+edi] punpcklbw mm2,mm7 punpcklbw mm3,mm7 paddw mm1,mm0 pmullw mm3,mm6 paddw mm1,mm2 pmullw mm1,mm5 add esi,4 psubusw mm3,mm1 psrlw mm3,6 packuswb mm3,mm3 dec ebp movd [edx-4],mm3 jne colloop_MMX ;************************** movd mm1,[esi+4] punpcklbw mm1,mm7 paddw mm0,mm1 movd mm1,[esi+edi*2+4] punpcklbw mm1,mm7 paddw mm0,mm1 movd mm1,[esi-8] punpcklbw mm1,mm7 psubw mm0,mm1 movd mm1,[esi+edi*2-8] punpcklbw mm1,mm7 psubw mm0,mm1 movd mm1,[esi+edi-4] punpcklbw mm1,mm7 movd mm2,[esi+edi+4] punpcklbw mm2,mm7 paddw mm1,mm2 paddw mm1,mm0 pmullw mm1,mm5 movd mm2,[esi+edi] punpcklbw mm2,mm7 pmullw mm2,mm6 psubusw mm2,mm1 psrlw mm2,6 packuswb mm2,mm2 movd [edx],mm2 add esi,4 add edx,4 dec ebp jne colloop_MMX ENDIF end