July 2001 - MPlayer-cvslog - lists.mplayerhq.hu

CVS: main/DOCS/Hungarian MPlayer-FAQ,1.23,1.24
by Winner of tha face compo 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/DOCS/Hungarian In directory mplayer:/var/tmp.root/cvs-serv1513/DOCS/Hungarian Modified Files: MPlayer-FAQ Log Message: Index: MPlayer-FAQ =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/Hungarian/MPlayer-FAQ,v retrieving revision 1.23 retrieving revision 1.24 diff -u -r1.23 -r1.24 --- MPlayer-FAQ 4 Jul 2001 20:18:24 -0000 1.23 +++ MPlayer-FAQ 6 Jul 2001 07:35:13 -0000 1.24 @@ -172,9 +172,13 @@ Xvideo image format: 0x32315659 (YV12) planar Xvideo image format: 0x30323449 (I420) planar ... -A: Lásd az elõzõ választ. +A1: Lásd az elõzõ választ. +A1: A 0.18pre4 verzió óta az MPlayer tartalmazza a libavcodec-et az ffmpeg + programból. Ez magában foglal egy C nyelvû DivX és OpenDivX dekódert. + A DivX dekóder YV12-es formátumot használ, szóval mûködni fog nálad. + Fordítsd be a libavcodec supportot. -Q: Ooo, mi az az "IdegCounter" ? +Q: Ööö, mi az az "IdegCounter" ? A: Az "Ideg" magyar szó, és a "Counter" (számláló) angol szó keveréke. Elõször A'rpi idegességét volt hivatott mérni, miután egy kis CVS kód (ooo) "rejtelyesen" eltûnt ;) @@ -262,3 +266,12 @@ film alatt! Kinek a hibája ez? A: Az nVidia driver-é. Ezek a hibák CSAK TNT/TNT2 kártyákkal jönnek elõ, és nem tehetünk róla, a hiba az Ön készülékében van. + +Q: Pontosan mi is ez a libavcodec? +A: Gerard Lantau ffmpeg programjábol van (http://ffmpeg.sourceforge.net) + C nyelvû DivX ;-) és OpenDivX codec-et tartalmaz. A tesztek szerint gyorsabb + mint a DirectShow codec (preprocessing nélkül), és csak pár százalékkal + lassabb mint a ProjectMayo OpenDivX dekódere. YV12-es pixelformátumot + használ, így a régi Voodoo tulajdonosok is használhatják YUV-ra. Továbbá + lehetõséget ad arra hogy minden gépen amin van C compiler, lehessen DivX-et + lejátszani. Egész jó kis cucc, nem? _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/DOCS MPlayer-FAQ,1.56,1.57
by Winner of tha face compo 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/DOCS In directory mplayer:/var/tmp.root/cvs-serv1513/DOCS Modified Files: MPlayer-FAQ Log Message: Index: MPlayer-FAQ =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/MPlayer-FAQ,v retrieving revision 1.56 retrieving revision 1.57 diff -u -r1.56 -r1.57 --- MPlayer-FAQ 4 Jul 2001 20:18:24 -0000 1.56 +++ MPlayer-FAQ 6 Jul 2001 07:35:13 -0000 1.57 @@ -169,7 +169,10 @@ Xvideo image format: 0x32315659 (YV12) planar Xvideo image format: 0x30323449 (I420) planar ... -A: See the previous answer. +A1: See the previous answer. +A2: Since 0.18pre4 we include libavcodec from the ffmpeg package. It contains + a C language DivX and OpenDivX decoder. The DivX decoder uses YV12 format + for output, thus it should work for you. Compile in libavcodec support. Q: Umm, what is "IdegCounter" ? A: A mixture of a hungarian and an english word. In english, "Ideg" means @@ -254,3 +257,12 @@ right under the movie! Whose fault is this? A: It's the nVidia X driver's. These bugs are ONLY with the TNT/TNT2 cards, and we can't do anything about it, it's not our bug. + +Q: What exactly is this libavcodec? +A: It's from Gerard Lantau's ffmpeg package (http://ffmpeg.sourceforge.net) + It has C language DivX ;-) and OpenDivX codec. Some benchmarks + showed that its DivX seems to be even faster than DirectShow codec without + preprocessing, and only a few percent slower than ProjectMayo's OpenDivX + decoder. It uses YV12 pixel format, so old Voodoo users can use it for YUV. + Also, open the possibility to play DivX on virtually every computer that + has a C compiler. Kinda great stuff, isn't it? _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/libavcodec/i386 fdct_mmx.s,NONE,1.1 fdctdata.c,NONE,1.1 sad_mmx.s,NONE,1.1
by Arpi of Ize 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/libavcodec/i386 In directory mplayer:/var/tmp.root/cvs-serv27291/i386 Added Files: fdct_mmx.s fdctdata.c sad_mmx.s Log Message: (modified) libavcodec from ffmpeg 0.4.3 --- NEW FILE --- ; ////////////////////////////////////////////////////////////////////////////// ; // ; // fdctam32.c - AP922 MMX(3D-Now) forward-DCT ; // ---------- ; // Intel Application Note AP-922 - fast, precise implementation of DCT ; // http://developer.intel.com/vtune/cbts/appnotes.htm ; // ---------- ; // ; // This routine can use a 3D-Now/MMX enhancement to increase the ; // accuracy of the fdct_col_4 macro. The dct_col function uses 3D-Now's ; // PMHULHRW instead of MMX's PMHULHW(and POR). The substitution improves ; // accuracy very slightly with performance penalty. If the target CPU ; // does not support 3D-Now, then this function cannot be executed. ; // ; // For a fast, precise MMX implementation of inverse-DCT ; // visit http://www.elecard.com/peter ; // ; // v1.0 07/22/2000 (initial release) ; // ; // liaor(a)iname.com http://members.tripod.com/~liaor ; ////////////////////////////////////////////////////////////////////////////// ;;; ;;; A.Stevens Jul 2000: ported to nasm syntax and disentangled from ;;; from Win**** compiler specific stuff. ;;; All the real work was done above though. ;;; See above for how to optimise quality on 3DNow! CPU's ;; ;; Macros for code-readability... ;; %define INP eax ; pointer to (short *blk) %define OUT ecx ; pointer to output (temporary store space qwTemp[]) %define TABLE ebx ; pointer to tab_frw_01234567[] %define TABLEF ebx ; pointer to tg_all_16 %define round_frw_row edx %define x0 INP + 0*16 %define x1 INP + 1*16 %define x2 INP + 2*16 %define x3 INP + 3*16 %define x4 INP + 4*16 %define x5 INP + 5*16 %define x6 INP + 6*16 %define x7 INP + 7*16 %define y0 OUT + 0*16 %define y1 OUT + 1*16 %define y2 OUT + 2*16 %define y3 OUT + 3*16 %define y4 OUT + 4*16 %define y5 OUT + 5*16 %define y6 OUT + 6*16 %define y7 OUT + 7*16 ;; ;; Constants for DCT ;; %define BITS_FRW_ACC 3 ; 2 or 3 for accuracy %define SHIFT_FRW_COL BITS_FRW_ACC %define SHIFT_FRW_ROW (BITS_FRW_ACC + 17) %define RND_FRW_ROW (1 << (SHIFT_FRW_ROW-1)) %define RND_FRW_COL (1 << (SHIFT_FRW_COL-1)) extern fdct_one_corr extern fdct_r_row ; Defined in C for convenience ;; ;; Concatenated table of forward dct transformation coeffs. ;; extern fdct_tg_all_16 ; Defined in C for convenience ;; Offsets into table.. %define tg_1_16 (TABLEF + 0) %define tg_2_16 (TABLEF + 8) %define tg_3_16 (TABLEF + 16) %define cos_4_16 (TABLEF + 24) %define ocos_4_16 (TABLEF + 32) ;; ;; Concatenated table of forward dct coefficients ;; extern tab_frw_01234567 ; Defined in C for convenience ;; Offsets into table.. SECTION .text global fdct_mmx ;;; ;;; void fdct_mmx( short *blk ) ;;; ; //////////////////////////////////////////////////////////////////////// ; // ; // The high-level pseudocode for the fdct_am32() routine : ; // ; // fdct_am32() ; // { ; // forward_dct_col03(); // dct_column transform on cols 0-3 ; // forward_dct_col47(); // dct_column transform on cols 4-7 ; // for ( j = 0; j < 8; j=j+1 ) ; // forward_dct_row1(j); // dct_row transform on row #j ; // } ; // ; align 32 fdct_mmx: push ebp ; save stack pointer mov ebp, esp ; link push ebx push ecx push edx push edi mov INP, [ebp+8]; ; input data is row 0 of blk[] ;// transform the left half of the matrix (4 columns) lea TABLEF, [fdct_tg_all_16]; mov OUT, INP; ; lea round_frw_col, [r_frw_col] ; for ( i = 0; i < 2; i = i + 1) ; the for-loop is executed twice. We are better off unrolling the ; loop to avoid branch misprediction. .mmx32_fdct_col03: movq mm0, [x1] ; 0 ; x1 ;; movq mm1, [x6] ; 1 ; x6 movq mm2, mm0 ; 2 ; x1 movq mm3, [x2] ; 3 ; x2 paddsw mm0, mm1 ; t1 = x[1] + x[6] movq mm4, [x5] ; 4 ; x5 psllw mm0, SHIFT_FRW_COL ; t1 movq mm5, [x0] ; 5 ; x0 paddsw mm4, mm3 ; t2 = x[2] + x[5] paddsw mm5, [x7] ; t0 = x[0] + x[7] psllw mm4, SHIFT_FRW_COL ; t2 movq mm6, mm0 ; 6 ; t1 psubsw mm2, mm1 ; 1 ; t6 = x[1] - x[6] movq mm1, [tg_2_16] ; 1 ; tg_2_16 psubsw mm0, mm4 ; tm12 = t1 - t2 movq mm7, [x3] ; 7 ; x3 pmulhw mm1, mm0 ; tm12*tg_2_16 paddsw mm7, [x4] ; t3 = x[3] + x[4] psllw mm5, SHIFT_FRW_COL ; t0 paddsw mm6, mm4 ; 4 ; tp12 = t1 + t2 psllw mm7, SHIFT_FRW_COL ; t3 movq mm4, mm5 ; 4 ; t0 psubsw mm5, mm7 ; tm03 = t0 - t3 paddsw mm1, mm5 ; y2 = tm03 + tm12*tg_2_16 paddsw mm4, mm7 ; 7 ; tp03 = t0 + t3 por mm1, [fdct_one_corr] ; correction y2 +0.5 psllw mm2, SHIFT_FRW_COL+1 ; t6 pmulhw mm5, [tg_2_16] ; tm03*tg_2_16 movq mm7, mm4 ; 7 ; tp03 psubsw mm3, [x5] ; t5 = x[2] - x[5] psubsw mm4, mm6 ; y4 = tp03 - tp12 movq [y2], mm1 ; 1 ; save y2 paddsw mm7, mm6 ; 6 ; y0 = tp03 + tp12 movq mm1, [x3] ; 1 ; x3 psllw mm3, SHIFT_FRW_COL+1 ; t5 psubsw mm1, [x4] ; t4 = x[3] - x[4] movq mm6, mm2 ; 6 ; t6 movq [y4], mm4 ; 4 ; save y4 paddsw mm2, mm3 ; t6 + t5 pmulhw mm2, [ocos_4_16] ; tp65 = (t6 + t5)*cos_4_16 psubsw mm6, mm3 ; 3 ; t6 - t5 pmulhw mm6, [ocos_4_16] ; tm65 = (t6 - t5)*cos_4_16 psubsw mm5, mm0 ; 0 ; y6 = tm03*tg_2_16 - tm12 por mm5, [fdct_one_corr] ; correction y6 +0.5 psllw mm1, SHIFT_FRW_COL ; t4 por mm2, [fdct_one_corr] ; correction tp65 +0.5 movq mm4, mm1 ; 4 ; t4 movq mm3, [x0] ; 3 ; x0 paddsw mm1, mm6 ; tp465 = t4 + tm65 psubsw mm3, [x7] ; t7 = x[0] - x[7] psubsw mm4, mm6 ; 6 ; tm465 = t4 - tm65 movq mm0, [tg_1_16] ; 0 ; tg_1_16 psllw mm3, SHIFT_FRW_COL ; t7 movq mm6, [tg_3_16] ; 6 ; tg_3_16 pmulhw mm0, mm1 ; tp465*tg_1_16 movq [y0], mm7 ; 7 ; save y0 pmulhw mm6, mm4 ; tm465*tg_3_16 movq [y6], mm5 ; 5 ; save y6 movq mm7, mm3 ; 7 ; t7 movq mm5, [tg_3_16] ; 5 ; tg_3_16 psubsw mm7, mm2 ; tm765 = t7 - tp65 paddsw mm3, mm2 ; 2 ; tp765 = t7 + tp65 pmulhw mm5, mm7 ; tm765*tg_3_16 paddsw mm0, mm3 ; y1 = tp765 + tp465*tg_1_16 paddsw mm6, mm4 ; tm465*tg_3_16 pmulhw mm3, [tg_1_16] ; tp765*tg_1_16 ;; por mm0, [fdct_one_corr] ; correction y1 +0.5 paddsw mm5, mm7 ; tm765*tg_3_16 psubsw mm7, mm6 ; 6 ; y3 = tm765 - tm465*tg_3_16 add INP, 0x08 ; ; increment pointer movq [y1], mm0 ; 0 ; save y1 paddsw mm5, mm4 ; 4 ; y5 = tm765*tg_3_16 + tm465 movq [y3], mm7 ; 7 ; save y3 psubsw mm3, mm1 ; 1 ; y7 = tp765*tg_1_16 - tp465 movq [y5], mm5 ; 5 ; save y5 .mmx32_fdct_col47: ; begin processing last four columns movq mm0, [x1] ; 0 ; x1 ;; movq [y7], mm3 ; 3 ; save y7 (columns 0-4) ;; movq mm1, [x6] ; 1 ; x6 movq mm2, mm0 ; 2 ; x1 movq mm3, [x2] ; 3 ; x2 paddsw mm0, mm1 ; t1 = x[1] + x[6] movq mm4, [x5] ; 4 ; x5 psllw mm0, SHIFT_FRW_COL ; t1 movq mm5, [x0] ; 5 ; x0 paddsw mm4, mm3 ; t2 = x[2] + x[5] paddsw mm5, [x7] ; t0 = x[0] + x[7] psllw mm4, SHIFT_FRW_COL ; t2 movq mm6, mm0 ; 6 ; t1 psubsw mm2, mm1 ; 1 ; t6 = x[1] - x[6] movq mm1, [tg_2_16] ; 1 ; tg_2_16 psubsw mm0, mm4 ; tm12 = t1 - t2 movq mm7, [x3] ; 7 ; x3 pmulhw mm1, mm0 ; tm12*tg_2_16 paddsw mm7, [x4] ; t3 = x[3] + x[4] psllw mm5, SHIFT_FRW_COL ; t0 paddsw mm6, mm4 ; 4 ; tp12 = t1 + t2 psllw mm7, SHIFT_FRW_COL ; t3 movq mm4, mm5 ; 4 ; t0 psubsw mm5, mm7 ; tm03 = t0 - t3 paddsw mm1, mm5 ; y2 = tm03 + tm12*tg_2_16 paddsw mm4, mm7 ; 7 ; tp03 = t0 + t3 por mm1, [fdct_one_corr] ; correction y2 +0.5 psllw mm2, SHIFT_FRW_COL+1 ; t6 pmulhw mm5, [tg_2_16] ; tm03*tg_2_16 movq mm7, mm4 ; 7 ; tp03 psubsw mm3, [x5] ; t5 = x[2] - x[5] psubsw mm4, mm6 ; y4 = tp03 - tp12 movq [y2+8], mm1 ; 1 ; save y2 paddsw mm7, mm6 ; 6 ; y0 = tp03 + tp12 movq mm1, [x3] ; 1 ; x3 psllw mm3, SHIFT_FRW_COL+1 ; t5 psubsw mm1, [x4] ; t4 = x[3] - x[4] movq mm6, mm2 ; 6 ; t6 movq [y4+8], mm4 ; 4 ; save y4 paddsw mm2, mm3 ; t6 + t5 pmulhw mm2, [ocos_4_16] ; tp65 = (t6 + t5)*cos_4_16 psubsw mm6, mm3 ; 3 ; t6 - t5 pmulhw mm6, [ocos_4_16] ; tm65 = (t6 - t5)*cos_4_16 psubsw mm5, mm0 ; 0 ; y6 = tm03*tg_2_16 - tm12 por mm5, [fdct_one_corr] ; correction y6 +0.5 psllw mm1, SHIFT_FRW_COL ; t4 por mm2, [fdct_one_corr] ; correction tp65 +0.5 movq mm4, mm1 ; 4 ; t4 movq mm3, [x0] ; 3 ; x0 paddsw mm1, mm6 ; tp465 = t4 + tm65 psubsw mm3, [x7] ; t7 = x[0] - x[7] psubsw mm4, mm6 ; 6 ; tm465 = t4 - tm65 movq mm0, [tg_1_16] ; 0 ; tg_1_16 psllw mm3, SHIFT_FRW_COL ; t7 movq mm6, [tg_3_16] ; 6 ; tg_3_16 pmulhw mm0, mm1 ; tp465*tg_1_16 movq [y0+8], mm7 ; 7 ; save y0 pmulhw mm6, mm4 ; tm465*tg_3_16 movq [y6+8], mm5 ; 5 ; save y6 movq mm7, mm3 ; 7 ; t7 movq mm5, [tg_3_16] ; 5 ; tg_3_16 psubsw mm7, mm2 ; tm765 = t7 - tp65 paddsw mm3, mm2 ; 2 ; tp765 = t7 + tp65 pmulhw mm5, mm7 ; tm765*tg_3_16 paddsw mm0, mm3 ; y1 = tp765 + tp465*tg_1_16 paddsw mm6, mm4 ; tm465*tg_3_16 pmulhw mm3, [tg_1_16] ; tp765*tg_1_16 ;; por mm0, [fdct_one_corr] ; correction y1 +0.5 paddsw mm5, mm7 ; tm765*tg_3_16 psubsw mm7, mm6 ; 6 ; y3 = tm765 - tm465*tg_3_16 ;; movq [y1+8], mm0 ; 0 ; save y1 paddsw mm5, mm4 ; 4 ; y5 = tm765*tg_3_16 + tm465 movq [y3+8], mm7 ; 7 ; save y3 psubsw mm3, mm1 ; 1 ; y7 = tp765*tg_1_16 - tp465 movq [y5+8], mm5 ; 5 ; save y5 movq [y7+8], mm3 ; 3 ; save y7 ; emms; ; } ; end of forward_dct_col07() ; done with dct_row transform ; fdct_mmx32_cols() -- ; the following subroutine repeats the row-transform operation, ; except with different shift&round constants. This version ; does NOT transpose the output again. Thus the final output ; is transposed with respect to the source. ; ; The output is stored into blk[], which destroys the original ; input data. mov INP, [ebp+8]; ;; row 0 mov edi, 0x08; ;x = 8 lea TABLE, [tab_frw_01234567]; ; row 0 mov OUT, INP; lea round_frw_row, [fdct_r_row]; ; for ( x = 8; x > 0; --x ) ; transform one row per iteration ; ---------- loop begin .lp_mmx_fdct_row1: movd mm5, [INP+12]; ; mm5 = 7 6 punpcklwd mm5, [INP+8] ; mm5 = 5 7 4 6 movq mm2, mm5; ; mm2 = 5 7 4 6 psrlq mm5, 32; ; mm5 = _ _ 5 7 movq mm0, [INP]; ; mm0 = 3 2 1 0 punpcklwd mm5, mm2;; mm5 = 4 5 6 7 movq mm1, mm0; ; mm1 = 3 2 1 0 paddsw mm0, mm5; ; mm0 = [3+4, 2+5, 1+6, 0+7] (xt3, xt2, xt1, xt0) psubsw mm1, mm5; ; mm1 = [3-4, 2-5, 1-6, 0-7] (xt7, xt6, xt5, xt4) movq mm2, mm0; ; mm2 = [ xt3 xt2 xt1 xt0 ] ;movq [ xt3xt2xt1xt0 ], mm0; ;movq [ xt7xt6xt5xt4 ], mm1; punpcklwd mm0, mm1;; mm0 = [ xt5 xt1 xt4 xt0 ] punpckhwd mm2, mm1;; mm2 = [ xt7 xt3 xt6 xt2 ] movq mm1, mm2; ; mm1 ;; shuffle bytes around ; movq mm0, [INP] ; 0 ; x3 x2 x1 x0 ; movq mm1, [INP+8] ; 1 ; x7 x6 x5 x4 movq mm2, mm0 ; 2 ; x3 x2 x1 x0 movq mm3, [TABLE] ; 3 ; w06 w04 w02 w00 punpcklwd mm0, mm1 ; x5 x1 x4 x0 movq mm5, mm0 ; 5 ; x5 x1 x4 x0 punpckldq mm0, mm0 ; x4 x0 x4 x0 [ xt2 xt0 xt2 xt0 ] movq mm4, [TABLE+8] ; 4 ; w07 w05 w03 w01 punpckhwd mm2, mm1 ; 1 ; x7 x3 x6 x2 pmaddwd mm3, mm0 ; x4*w06+x0*w04 x4*w02+x0*w00 movq mm6, mm2 ; 6 ; x7 x3 x6 x2 movq mm1, [TABLE+32] ; 1 ; w22 w20 w18 w16 punpckldq mm2, mm2 ; x6 x2 x6 x2 [ xt3 xt1 xt3 xt1 ] pmaddwd mm4, mm2 ; x6*w07+x2*w05 x6*w03+x2*w01 punpckhdq mm5, mm5 ; x5 x1 x5 x1 [ xt6 xt4 xt6 xt4 ] pmaddwd mm0, [TABLE+16] ; x4*w14+x0*w12 x4*w10+x0*w08 punpckhdq mm6, mm6 ; x7 x3 x7 x3 [ xt7 xt5 xt7 xt5 ] movq mm7, [TABLE+40] ; 7 ; w23 w21 w19 w17 pmaddwd mm1, mm5 ; x5*w22+x1*w20 x5*w18+x1*w16 ;mm3 = a1, a0 (y2,y0) ;mm1 = b1, b0 (y3,y1) ;mm0 = a3,a2 (y6,y4) ;mm5 = b3,b2 (y7,y5) paddd mm3, [round_frw_row] ; +rounder (y2,y0) pmaddwd mm7, mm6 ; x7*w23+x3*w21 x7*w19+x3*w17 pmaddwd mm2, [TABLE+24] ; x6*w15+x2*w13 x6*w11+x2*w09 paddd mm3, mm4 ; 4 ; a1=sum(even1) a0=sum(even0) ; now ( y2, y0) pmaddwd mm5, [TABLE+48] ; x5*w30+x1*w28 x5*w26+x1*w24 ;; pmaddwd mm6, [TABLE+56] ; x7*w31+x3*w29 x7*w27+x3*w25 paddd mm1, mm7 ; 7 ; b1=sum(odd1) b0=sum(odd0) ; now ( y3, y1) paddd mm0, [round_frw_row] ; +rounder (y6,y4) psrad mm3, SHIFT_FRW_ROW ; (y2, y0) paddd mm1, [round_frw_row] ; +rounder (y3,y1) paddd mm0, mm2 ; 2 ; a3=sum(even3) a2=sum(even2) ; now (y6, y4) paddd mm5, [round_frw_row] ; +rounder (y7,y5) psrad mm1, SHIFT_FRW_ROW ; y1=a1+b1 y0=a0+b0 paddd mm5, mm6 ; 6 ; b3=sum(odd3) b2=sum(odd2) ; now ( y7, y5) psrad mm0, SHIFT_FRW_ROW ;y3=a3+b3 y2=a2+b2 add OUT, 16; ; increment row-output address by 1 row psrad mm5, SHIFT_FRW_ROW ; y4=a3-b3 y5=a2-b2 add INP, 16; ; increment row-address by 1 row packssdw mm3, mm0 ; 0 ; y6 y4 y2 y0 packssdw mm1, mm5 ; 3 ; y7 y5 y3 y1 movq mm6, mm3; ; mm0 = y6 y4 y2 y0 punpcklwd mm3, mm1; ; y3 y2 y1 y0 sub edi, 0x01; ; i = i - 1 punpckhwd mm6, mm1; ; y7 y6 y5 y4 add TABLE,64; ; increment to next table movq [OUT-16], mm3 ; 1 ; save y3 y2 y1 y0 movq [OUT-8], mm6 ; 7 ; save y7 y6 y5 y4 cmp edi, 0x00; jg near .lp_mmx_fdct_row1; ; begin fdct processing on next row ;; ;; Tidy up and return ;; pop edi pop edx pop ecx pop ebx pop ebp ; restore stack pointer emms ret --- NEW FILE --- ////////////////////////////////////////////////////////////////////////////// // // fdctam32.c - AP922 MMX(3D-Now) forward-DCT // ---------- // Intel Application Note AP-922 - fast, precise implementation of DCT // http://developer.intel.com/vtune/cbts/appnotes.htm // ---------- // // This routine uses a 3D-Now/MMX enhancement to increase the // accuracy of the fdct_col_4 macro. The dct_col function uses 3D-Now's // PMHULHRW instead of MMX's PMHULHW(and POR). The substitution improves // accuracy very slightly with performance penalty. If the target CPU // does not support 3D-Now, then this function cannot be executed. // fdctmm32.c contains the standard MMX implementation of AP-922. // // For a fast, precise MMX implementation of inverse-DCT // visit http://www.elecard.com/peter // // v1.0 07/22/2000 (initial release) // Initial release of AP922 MMX(3D-Now) forward_DCT. // This code was tested with Visual C++ 6.0Pro + service_pack4 + // processor_pack_beta! If you have the processor_pack_beta, you can // remove the #include for amd3dx.h, and substitute the 'normal' // assembly lines for the macro'd versions. Otherwise, this // code should compile 'as is', under Visual C++ 6.0 Pro. // // liaor(a)iname.com http://members.tripod.com/~liaor ////////////////////////////////////////////////////////////////////////////// #include <inttypes.h> ////////////////////////////////////////////////////////////////////// // // constants for the forward DCT // ----------------------------- // // Be sure to check that your compiler is aligning all constants to QWORD // (8-byte) memory boundaries! Otherwise the unaligned memory access will // severely stall MMX execution. // ////////////////////////////////////////////////////////////////////// #define BITS_FRW_ACC 3 //; 2 or 3 for accuracy #define SHIFT_FRW_COL BITS_FRW_ACC #define SHIFT_FRW_ROW (BITS_FRW_ACC + 17) //#define RND_FRW_ROW (262144 * (BITS_FRW_ACC - 1)) //; 1 << (SHIFT_FRW_ROW-1) #define RND_FRW_ROW (1 << (SHIFT_FRW_ROW-1)) //#define RND_FRW_COL (2 * (BITS_FRW_ACC - 1)) //; 1 << (SHIFT_FRW_COL-1) #define RND_FRW_COL (1 << (SHIFT_FRW_COL-1)) //concatenated table, for forward DCT transformation const int16_t fdct_tg_all_16[] = { 13036, 13036, 13036, 13036, // tg * (2<<16) + 0.5 27146, 27146, 27146, 27146, // tg * (2<<16) + 0.5 -21746, -21746, -21746, -21746, // tg * (2<<16) + 0.5 -19195, -19195, -19195, -19195, //cos * (2<<16) + 0.5 23170, 23170, 23170, 23170 }; //cos * (2<<15) + 0.5 const long long fdct_one_corr = 0x0001000100010001LL; const long fdct_r_row[2] = {RND_FRW_ROW, RND_FRW_ROW }; const int16_t tab_frw_01234567[] = { // forward_dct coeff table //row0 16384, 16384, 21407, -8867, // w09 w01 w08 w00 16384, 16384, 8867, -21407, // w13 w05 w12 w04 16384, -16384, 8867, 21407, // w11 w03 w10 w02 -16384, 16384, -21407, -8867, // w15 w07 w14 w06 22725, 12873, 19266, -22725, // w22 w20 w18 w16 19266, 4520, -4520, -12873, // w23 w21 w19 w17 12873, 4520, 4520, 19266, // w30 w28 w26 w24 -22725, 19266, -12873, -22725, // w31 w29 w27 w25 //row1 22725, 22725, 29692, -12299, // w09 w01 w08 w00 22725, 22725, 12299, -29692, // w13 w05 w12 w04 22725, -22725, 12299, 29692, // w11 w03 w10 w02 -22725, 22725, -29692, -12299, // w15 w07 w14 w06 31521, 17855, 26722, -31521, // w22 w20 w18 w16 26722, 6270, -6270, -17855, // w23 w21 w19 w17 17855, 6270, 6270, 26722, // w30 w28 w26 w24 -31521, 26722, -17855, -31521, // w31 w29 w27 w25 //row2 21407, 21407, 27969, -11585, // w09 w01 w08 w00 21407, 21407, 11585, -27969, // w13 w05 w12 w04 21407, -21407, 11585, 27969, // w11 w03 w10 w02 -21407, 21407, -27969, -11585, // w15 w07 w14 w06 29692, 16819, 25172, -29692, // w22 w20 w18 w16 25172, 5906, -5906, -16819, // w23 w21 w19 w17 16819, 5906, 5906, 25172, // w30 w28 w26 w24 -29692, 25172, -16819, -29692, // w31 w29 w27 w25 //row3 19266, 19266, 25172, -10426, // w09 w01 w08 w00 19266, 19266, 10426, -25172, // w13 w05 w12 w04 19266, -19266, 10426, 25172, // w11 w03 w10 w02 -19266, 19266, -25172, -10426, // w15 w07 w14 w06, 26722, 15137, 22654, -26722, // w22 w20 w18 w16 22654, 5315, -5315, -15137, // w23 w21 w19 w17 15137, 5315, 5315, 22654, // w30 w28 w26 w24 -26722, 22654, -15137, -26722, // w31 w29 w27 w25, //row4 16384, 16384, 21407, -8867, // w09 w01 w08 w00 16384, 16384, 8867, -21407, // w13 w05 w12 w04 16384, -16384, 8867, 21407, // w11 w03 w10 w02 -16384, 16384, -21407, -8867, // w15 w07 w14 w06 22725, 12873, 19266, -22725, // w22 w20 w18 w16 19266, 4520, -4520, -12873, // w23 w21 w19 w17 12873, 4520, 4520, 19266, // w30 w28 w26 w24 -22725, 19266, -12873, -22725, // w31 w29 w27 w25 //row5 19266, 19266, 25172, -10426, // w09 w01 w08 w00 19266, 19266, 10426, -25172, // w13 w05 w12 w04 19266, -19266, 10426, 25172, // w11 w03 w10 w02 -19266, 19266, -25172, -10426, // w15 w07 w14 w06 26722, 15137, 22654, -26722, // w22 w20 w18 w16 22654, 5315, -5315, -15137, // w23 w21 w19 w17 15137, 5315, 5315, 22654, // w30 w28 w26 w24 -26722, 22654, -15137, -26722, // w31 w29 w27 w25 //row6 21407, 21407, 27969, -11585, // w09 w01 w08 w00 21407, 21407, 11585, -27969, // w13 w05 w12 w04 21407, -21407, 11585, 27969, // w11 w03 w10 w02 -21407, 21407, -27969, -11585, // w15 w07 w14 w06, 29692, 16819, 25172, -29692, // w22 w20 w18 w16 25172, 5906, -5906, -16819, // w23 w21 w19 w17 16819, 5906, 5906, 25172, // w30 w28 w26 w24 -29692, 25172, -16819, -29692, // w31 w29 w27 w25, //row7 22725, 22725, 29692, -12299, // w09 w01 w08 w00 22725, 22725, 12299, -29692, // w13 w05 w12 w04 22725, -22725, 12299, 29692, // w11 w03 w10 w02 -22725, 22725, -29692, -12299, // w15 w07 w14 w06, 31521, 17855, 26722, -31521, // w22 w20 w18 w16 26722, 6270, -6270, -17855, // w23 w21 w19 w17 17855, 6270, 6270, 26722, // w30 w28 w26 w24 -31521, 26722, -17855, -31521 // w31 w29 w27 w25 }; --- NEW FILE --- ; MMX/SSE optimized routines for SAD of 16*16 macroblocks ; Copyright (C) Juan J. Sierralta P. <juanjo(a)atmlab.utfsm.cl> ; ; dist1_* Original Copyright (C) 2000 Chris Atenasio <chris(a)crud.net> ; Enhancements and rest Copyright (C) 2000 Andrew Stevens <as(a)comlab.ox.ac.uk> ; ; 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. ; global pix_abs16x16_mmx ; int pix_abs16x16_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h); ; esi = p1 (init: blk1) ; edi = p2 (init: blk2) ; ecx = rowsleft (init: h) ; edx = lx; ; mm0 = distance accumulators (4 words) ; mm1 = distance accumulators (4 words) ; mm2 = temp ; mm3 = temp ; mm4 = temp ; mm5 = temp ; mm6 = 0 ; mm7 = temp align 32 pix_abs16x16_mmx: push ebp ; save frame pointer mov ebp, esp push ebx ; Saves registers (called saves convention in push ecx ; x86 GCC it seems) push edx ; push esi push edi pxor mm0, mm0 ; zero acculumators pxor mm1, mm1 pxor mm6, mm6 mov esi, [ebp+8] ; get pix1 mov edi, [ebp+12] ; get pix2 mov edx, [ebp+16] ; get lx mov ecx, [ebp+20] ; get rowsleft jmp .nextrow align 32 .nextrow: ; First 8 bytes of the row movq mm4, [edi] ; load first 8 bytes of pix2 row movq mm5, [esi] ; load first 8 bytes of pix1 row movq mm3, mm4 ; mm4 := abs(mm4-mm5) movq mm2,[esi+8] ; load last 8 bytes of pix1 row psubusb mm4, mm5 movq mm7,[edi+8] ; load last 8 bytes of pix2 row psubusb mm5, mm3 por mm4, mm5 ; Last 8 bytes of the row movq mm3, mm7 ; mm7 := abs(mm7-mm2) psubusb mm7, mm2 psubusb mm2, mm3 por mm7, mm2 ; Now mm4 and mm7 have 16 absdiffs to add ; First 8 bytes of the row2 add edi, edx movq mm2, [edi] ; load first 8 bytes of pix2 row add esi, edx movq mm5, [esi] ; load first 8 bytes of pix1 row movq mm3, mm2 ; mm2 := abs(mm2-mm5) psubusb mm2, mm5 movq mm6,[esi+8] ; load last 8 bytes of pix1 row psubusb mm5, mm3 por mm2, mm5 ; Last 8 bytes of the row2 movq mm5,[edi+8] ; load last 8 bytes of pix2 row movq mm3, mm5 ; mm5 := abs(mm5-mm6) psubusb mm5, mm6 psubusb mm6, mm3 por mm5, mm6 ; Now mm2, mm4, mm5, mm7 have 32 absdiffs movq mm3, mm7 pxor mm6, mm6 ; Zero mm6 punpcklbw mm3, mm6 ; Unpack to words and add punpckhbw mm7, mm6 paddusw mm7, mm3 movq mm3, mm5 punpcklbw mm3, mm6 ; Unpack to words and add punpckhbw mm5, mm6 paddusw mm5, mm3 paddusw mm0, mm7 ; Add to the acumulator (mm0) paddusw mm1, mm5 ; Add to the acumulator (mm1) movq mm3, mm4 punpcklbw mm3, mm6 ; Unpack to words and add punpckhbw mm4, mm6 movq mm5, mm2 paddusw mm4, mm3 punpcklbw mm5, mm6 ; Unpack to words and add punpckhbw mm2, mm6 paddusw mm2, mm5 ; Loop termination add esi, edx ; update pointers to next row paddusw mm0, mm4 ; Add to the acumulator (mm0) add edi, edx sub ecx,2 paddusw mm1, mm2 ; Add to the acumulator (mm1) test ecx, ecx ; check rowsleft jnz near .nextrow paddusw mm0, mm1 movq mm2, mm0 ; Copy mm0 to mm2 psrlq mm2, 32 paddusw mm0, mm2 ; Add movq mm3, mm0 psrlq mm3, 16 paddusw mm0, mm3 movd eax, mm0 ; Store return value and eax, 0xffff pop edi pop esi pop edx pop ecx pop ebx pop ebp ; restore stack pointer ;emms ; clear mmx registers ret ; return global pix_abs16x16_sse ; int pix_abs16x16_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h); ; esi = p1 (init: blk1) ; edi = p2 (init: blk2) ; ecx = rowsleft (init: h) ; edx = lx; ; mm0 = distance accumulators (4 words) ; mm1 = distance accumulators (4 words) ; mm2 = temp ; mm3 = temp ; mm4 = temp ; mm5 = temp ; mm6 = temp ; mm7 = temp align 32 pix_abs16x16_sse: push ebp ; save frame pointer mov ebp, esp push ebx ; Saves registers (called saves convention in push ecx ; x86 GCC it seems) push edx ; push esi push edi pxor mm0, mm0 ; zero acculumators pxor mm1, mm1 mov esi, [ebp+8] ; get pix1 mov edi, [ebp+12] ; get pix2 mov edx, [ebp+16] ; get lx mov ecx, [ebp+20] ; get rowsleft jmp .next4row align 32 .next4row: ; First row movq mm4, [edi] ; load first 8 bytes of pix2 row movq mm5, [edi+8] ; load last 8 bytes of pix2 row psadbw mm4, [esi] ; SAD of first 8 bytes psadbw mm5, [esi+8] ; SAD of last 8 bytes paddw mm0, mm4 ; Add to acumulators paddw mm1, mm5 ; Second row add edi, edx; add esi, edx; movq mm6, [edi] ; load first 8 bytes of pix2 row movq mm7, [edi+8] ; load last 8 bytes of pix2 row psadbw mm6, [esi] ; SAD of first 8 bytes psadbw mm7, [esi+8] ; SAD of last 8 bytes paddw mm0, mm6 ; Add to acumulators paddw mm1, mm7 ; Third row add edi, edx; add esi, edx; movq mm4, [edi] ; load first 8 bytes of pix2 row movq mm5, [edi+8] ; load last 8 bytes of pix2 row psadbw mm4, [esi] ; SAD of first 8 bytes psadbw mm5, [esi+8] ; SAD of last 8 bytes paddw mm0, mm4 ; Add to acumulators paddw mm1, mm5 ; Fourth row add edi, edx; add esi, edx; movq mm6, [edi] ; load first 8 bytes of pix2 row movq mm7, [edi+8] ; load last 8 bytes of pix2 row psadbw mm6, [esi] ; SAD of first 8 bytes psadbw mm7, [esi+8] ; SAD of last 8 bytes paddw mm0, mm6 ; Add to acumulators paddw mm1, mm7 ; Loop termination add esi, edx ; update pointers to next row add edi, edx sub ecx,4 test ecx, ecx ; check rowsleft jnz near .next4row paddd mm0, mm1 ; Sum acumulators movd eax, mm0 ; Store return value pop edi pop esi pop edx pop ecx pop ebx pop ebp ; restore stack pointer ;emms ; clear mmx registers ret ; return global pix_abs16x16_x2_mmx ; int pix_abs16x16_x2_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h); ; esi = p1 (init: blk1) ; edi = p2 (init: blk2) ; ecx = rowsleft (init: h) ; edx = lx; ; mm0 = distance accumulators (4 words) ; mm1 = distance accumulators (4 words) ; mm2 = temp ; mm3 = temp ; mm4 = temp ; mm5 = temp ; mm6 = 0 ; mm7 = temp align 32 pix_abs16x16_x2_mmx: push ebp ; save frame pointer mov ebp, esp push ebx ; Saves registers (called saves convention in push ecx ; x86 GCC it seems) push edx ; push esi push edi pxor mm0, mm0 ; zero acculumators pxor mm1, mm1 pxor mm6, mm6 mov esi, [ebp+8] ; get pix1 mov edi, [ebp+12] ; get pix2 mov edx, [ebp+16] ; get lx mov ecx, [ebp+20] ; get rowsleft jmp .nextrow_x2 align 32 .nextrow_x2: ; First 8 bytes of the row movq mm4, [edi] ; load first 8 bytes of pix2 row movq mm5, [edi+1] ; load bytes 1-8 of pix2 row movq mm2, mm4 ; copy mm4 on mm2 movq mm3, mm5 ; copy mm5 on mm3 punpcklbw mm4, mm6 ; first 4 bytes of [edi] on mm4 punpcklbw mm5, mm6 ; first 4 bytes of [edi+1] on mm5 paddusw mm4, mm5 ; mm4 := first 4 bytes interpolated in words psrlw mm4, 1 punpckhbw mm2, mm6 ; last 4 bytes of [edi] on mm2 punpckhbw mm3, mm6 ; last 4 bytes of [edi+1] on mm3 paddusw mm2, mm3 ; mm2 := last 4 bytes interpolated in words psrlw mm2, 1 packuswb mm4, mm2 ; pack 8 bytes interpolated on mm4 movq mm5,[esi] ; load first 8 bytes of pix1 row movq mm3, mm4 ; mm4 := abs(mm4-mm5) psubusb mm4, mm5 psubusb mm5, mm3 por mm4, mm5 ; Last 8 bytes of the row movq mm7, [edi+8] ; load last 8 bytes of pix2 row movq mm5, [edi+9] ; load bytes 10-17 of pix2 row movq mm2, mm7 ; copy mm7 on mm2 movq mm3, mm5 ; copy mm5 on mm3 punpcklbw mm7, mm6 ; first 4 bytes of [edi+8] on mm7 punpcklbw mm5, mm6 ; first 4 bytes of [edi+9] on mm5 paddusw mm7, mm5 ; mm1 := first 4 bytes interpolated in words psrlw mm7, 1 punpckhbw mm2, mm6 ; last 4 bytes of [edi] on mm2 punpckhbw mm3, mm6 ; last 4 bytes of [edi+1] on mm3 paddusw mm2, mm3 ; mm2 := last 4 bytes interpolated in words psrlw mm2, 1 packuswb mm7, mm2 ; pack 8 bytes interpolated on mm1 movq mm5,[esi+8] ; load last 8 bytes of pix1 row movq mm3, mm7 ; mm7 := abs(mm1-mm5) psubusb mm7, mm5 psubusb mm5, mm3 por mm7, mm5 ; Now mm4 and mm7 have 16 absdiffs to add movq mm3, mm4 ; Make copies of these bytes movq mm2, mm7 punpcklbw mm4, mm6 ; Unpack to words and add punpcklbw mm7, mm6 paddusw mm4, mm7 paddusw mm0, mm4 ; Add to the acumulator (mm0) punpckhbw mm3, mm6 ; Unpack to words and add punpckhbw mm2, mm6 paddusw mm3, mm2 paddusw mm1, mm3 ; Add to the acumulator (mm1) ; Loop termination add esi, edx ; update pointers to next row add edi, edx sub ecx,1 test ecx, ecx ; check rowsleft jnz near .nextrow_x2 paddusw mm0, mm1 movq mm1, mm0 ; Copy mm0 to mm1 psrlq mm1, 32 paddusw mm0, mm1 ; Add movq mm2, mm0 psrlq mm2, 16 paddusw mm0, mm2 movd eax, mm0 ; Store return value and eax, 0xffff pop edi pop esi pop edx pop ecx pop ebx pop ebp ; restore stack pointer emms ; clear mmx registers ret ; return global pix_abs16x16_y2_mmx ; int pix_abs16x16_y2_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h); ; esi = p1 (init: blk1) ; edi = p2 (init: blk2) ; ebx = p2 + lx ; ecx = rowsleft (init: h) ; edx = lx; ; mm0 = distance accumulators (4 words) ; mm1 = distance accumulators (4 words) ; mm2 = temp ; mm3 = temp ; mm4 = temp ; mm5 = temp ; mm6 = 0 ; mm7 = temp align 32 pix_abs16x16_y2_mmx: push ebp ; save frame pointer mov ebp, esp push ebx ; Saves registers (called saves convention in push ecx ; x86 GCC it seems) push edx ; push esi push edi pxor mm0, mm0 ; zero acculumators pxor mm1, mm1 pxor mm6, mm6 mov esi, [ebp+8] ; get pix1 mov edi, [ebp+12] ; get pix2 mov edx, [ebp+16] ; get lx mov ecx, [ebp+20] ; get rowsleft mov ebx, edi add ebx, edx jmp .nextrow_y2 align 32 .nextrow_y2: ; First 8 bytes of the row movq mm4, [edi] ; load first 8 bytes of pix2 row movq mm5, [ebx] ; load bytes 1-8 of pix2 row movq mm2, mm4 ; copy mm4 on mm2 movq mm3, mm5 ; copy mm5 on mm3 punpcklbw mm4, mm6 ; first 4 bytes of [edi] on mm4 punpcklbw mm5, mm6 ; first 4 bytes of [ebx] on mm5 paddusw mm4, mm5 ; mm4 := first 4 bytes interpolated in words psrlw mm4, 1 punpckhbw mm2, mm6 ; last 4 bytes of [edi] on mm2 punpckhbw mm3, mm6 ; last 4 bytes of [edi+1] on mm3 paddusw mm2, mm3 ; mm2 := last 4 bytes interpolated in words psrlw mm2, 1 packuswb mm4, mm2 ; pack 8 bytes interpolated on mm4 movq mm5,[esi] ; load first 8 bytes of pix1 row movq mm3, mm4 ; mm4 := abs(mm4-mm5) psubusb mm4, mm5 psubusb mm5, mm3 por mm4, mm5 ; Last 8 bytes of the row movq mm7, [edi+8] ; load last 8 bytes of pix2 row movq mm5, [ebx+8] ; load bytes 10-17 of pix2 row movq mm2, mm7 ; copy mm7 on mm2 movq mm3, mm5 ; copy mm5 on mm3 punpcklbw mm7, mm6 ; first 4 bytes of [edi+8] on mm7 punpcklbw mm5, mm6 ; first 4 bytes of [ebx+8] on mm5 paddusw mm7, mm5 ; mm1 := first 4 bytes interpolated in words psrlw mm7, 1 punpckhbw mm2, mm6 ; last 4 bytes of [edi+8] on mm2 punpckhbw mm3, mm6 ; last 4 bytes of [ebx+8] on mm3 paddusw mm2, mm3 ; mm2 := last 4 bytes interpolated in words psrlw mm2, 1 packuswb mm7, mm2 ; pack 8 bytes interpolated on mm1 movq mm5,[esi+8] ; load last 8 bytes of pix1 row movq mm3, mm7 ; mm7 := abs(mm1-mm5) psubusb mm7, mm5 psubusb mm5, mm3 por mm7, mm5 ; Now mm4 and mm7 have 16 absdiffs to add movq mm3, mm4 ; Make copies of these bytes movq mm2, mm7 punpcklbw mm4, mm6 ; Unpack to words and add punpcklbw mm7, mm6 paddusw mm4, mm7 paddusw mm0, mm4 ; Add to the acumulator (mm0) punpckhbw mm3, mm6 ; Unpack to words and add punpckhbw mm2, mm6 paddusw mm3, mm2 paddusw mm1, mm3 ; Add to the acumulator (mm1) ; Loop termination add esi, edx ; update pointers to next row add edi, edx add ebx, edx sub ecx,1 test ecx, ecx ; check rowsleft jnz near .nextrow_y2 paddusw mm0, mm1 movq mm1, mm0 ; Copy mm0 to mm1 psrlq mm1, 32 paddusw mm0, mm1 ; Add movq mm2, mm0 psrlq mm2, 16 paddusw mm0, mm2 movd eax, mm0 ; Store return value and eax, 0xffff pop edi pop esi pop edx pop ecx pop ebx pop ebp ; restore stack pointer emms ; clear mmx registers ret ; return global pix_abs16x16_xy2_mmx ; int pix_abs16x16_xy2_mmx(unsigned char *p1,unsigned char *p2,int lx,int h); ; esi = p1 (init: blk1) ; edi = p2 (init: blk2) ; ebx = p1+lx ; ecx = rowsleft (init: h) ; edx = lx; ; mm0 = distance accumulators (4 words) ; mm1 = bytes p2 ; mm2 = bytes p1 ; mm3 = bytes p1+lx ; I'd love to find someplace to stash p1+1 and p1+lx+1's bytes ; but I don't think thats going to happen in iA32-land... ; mm4 = temp 4 bytes in words interpolating p1, p1+1 ; mm5 = temp 4 bytes in words from p2 ; mm6 = temp comparison bit mask p1,p2 ; mm7 = temp comparison bit mask p2,p1 align 32 pix_abs16x16_xy2_mmx: push ebp ; save stack pointer mov ebp, esp ; so that we can do this push ebx ; Saves registers (called saves convention in push ecx ; x86 GCC it seems) push edx ; push esi push edi pxor mm0, mm0 ; zero acculumators mov esi, [ebp+12] ; get p1 mov edi, [ebp+8] ; get p2 mov edx, [ebp+16] ; get lx mov ecx, [ebp+20] ; rowsleft := h mov ebx, esi add ebx, edx jmp .nextrowmm11 ; snap to it align 32 .nextrowmm11: ;; ;; First 8 bytes of row ;; ;; First 4 bytes of 8 movq mm4, [esi] ; mm4 := first 4 bytes p1 pxor mm7, mm7 movq mm2, mm4 ; mm2 records all 8 bytes punpcklbw mm4, mm7 ; First 4 bytes p1 in Words... movq mm6, [ebx] ; mm6 := first 4 bytes p1+lx movq mm3, mm6 ; mm3 records all 8 bytes punpcklbw mm6, mm7 paddw mm4, mm6 movq mm5, [esi+1] ; mm5 := first 4 bytes p1+1 punpcklbw mm5, mm7 ; First 4 bytes p1 in Words... paddw mm4, mm5 movq mm6, [ebx+1] ; mm6 := first 4 bytes p1+lx+1 punpcklbw mm6, mm7 paddw mm4, mm6 psrlw mm4, 2 ; mm4 := First 4 bytes interpolated in words movq mm5, [edi] ; mm5:=first 4 bytes of p2 in words movq mm1, mm5 punpcklbw mm5, mm7 movq mm7,mm4 pcmpgtw mm7,mm5 ; mm7 := [i : W0..3,mm4>mm5] movq mm6,mm4 ; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)] psubw mm6,mm5 pand mm6, mm7 paddw mm0, mm6 ; Add to accumulator movq mm6,mm5 ; mm6 := [i : W0..3,mm5>mm4] pcmpgtw mm6,mm4 psubw mm5,mm4 ; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)] pand mm5, mm6 paddw mm0, mm5 ; Add to accumulator ;; Second 4 bytes of 8 movq mm4, mm2 ; mm4 := Second 4 bytes p1 in words pxor mm7, mm7 punpckhbw mm4, mm7 movq mm6, mm3 ; mm6 := Second 4 bytes p1+1 in words punpckhbw mm6, mm7 paddw mm4, mm6 movq mm5, [esi+1] ; mm5 := first 4 bytes p1+1 punpckhbw mm5, mm7 ; First 4 bytes p1 in Words... paddw mm4, mm5 movq mm6, [ebx+1] ; mm6 := first 4 bytes p1+lx+1 punpckhbw mm6, mm7 paddw mm4, mm6 psrlw mm4, 2 ; mm4 := First 4 bytes interpolated in words movq mm5, mm1 ; mm5:= second 4 bytes of p2 in words punpckhbw mm5, mm7 movq mm7,mm4 pcmpgtw mm7,mm5 ; mm7 := [i : W0..3,mm4>mm5] movq mm6,mm4 ; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)] psubw mm6,mm5 pand mm6, mm7 paddw mm0, mm6 ; Add to accumulator movq mm6,mm5 ; mm6 := [i : W0..3,mm5>mm4] pcmpgtw mm6,mm4 psubw mm5,mm4 ; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)] pand mm5, mm6 paddw mm0, mm5 ; Add to accumulator ;; ;; Second 8 bytes of row ;; ;; First 4 bytes of 8 movq mm4, [esi+8] ; mm4 := first 4 bytes p1+8 pxor mm7, mm7 movq mm2, mm4 ; mm2 records all 8 bytes punpcklbw mm4, mm7 ; First 4 bytes p1 in Words... movq mm6, [ebx+8] ; mm6 := first 4 bytes p1+lx+8 movq mm3, mm6 ; mm3 records all 8 bytes punpcklbw mm6, mm7 paddw mm4, mm6 movq mm5, [esi+9] ; mm5 := first 4 bytes p1+9 punpcklbw mm5, mm7 ; First 4 bytes p1 in Words... paddw mm4, mm5 movq mm6, [ebx+9] ; mm6 := first 4 bytes p1+lx+9 punpcklbw mm6, mm7 paddw mm4, mm6 psrlw mm4, 2 ; mm4 := First 4 bytes interpolated in words movq mm5, [edi+8] ; mm5:=first 4 bytes of p2+8 in words movq mm1, mm5 punpcklbw mm5, mm7 movq mm7,mm4 pcmpgtw mm7,mm5 ; mm7 := [i : W0..3,mm4>mm5] movq mm6,mm4 ; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)] psubw mm6,mm5 pand mm6, mm7 paddw mm0, mm6 ; Add to accumulator movq mm6,mm5 ; mm6 := [i : W0..3,mm5>mm4] pcmpgtw mm6,mm4 psubw mm5,mm4 ; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)] pand mm5, mm6 paddw mm0, mm5 ; Add to accumulator ;; Second 4 bytes of 8 movq mm4, mm2 ; mm4 := Second 4 bytes p1 in words pxor mm7, mm7 punpckhbw mm4, mm7 movq mm6, mm3 ; mm6 := Second 4 bytes p1+1 in words punpckhbw mm6, mm7 paddw mm4, mm6 movq mm5, [esi+9] ; mm5 := first 4 bytes p1+1 punpckhbw mm5, mm7 ; First 4 bytes p1 in Words... paddw mm4, mm5 movq mm6, [ebx+9] ; mm6 := first 4 bytes p1+lx+1 punpckhbw mm6, mm7 paddw mm4, mm6 psrlw mm4, 2 ; mm4 := First 4 bytes interpolated in words movq mm5, mm1 ; mm5:= second 4 bytes of p2 in words punpckhbw mm5, mm7 movq mm7,mm4 pcmpgtw mm7,mm5 ; mm7 := [i : W0..3,mm4>mm5] movq mm6,mm4 ; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)] psubw mm6,mm5 pand mm6, mm7 paddw mm0, mm6 ; Add to accumulator movq mm6,mm5 ; mm6 := [i : W0..3,mm5>mm4] pcmpgtw mm6,mm4 psubw mm5,mm4 ; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)] pand mm5, mm6 paddw mm0, mm5 ; Add to accumulator ;; ;; Loop termination condition... and stepping ;; add esi, edx ; update pointer to next row add edi, edx ; ditto add ebx, edx sub ecx,1 test ecx, ecx ; check rowsleft jnz near .nextrowmm11 ;; Sum the Accumulators movq mm4, mm0 psrlq mm4, 32 paddw mm0, mm4 movq mm6, mm0 psrlq mm6, 16 paddw mm0, mm6 movd eax, mm0 ; store return value and eax, 0xffff pop edi pop esi pop edx pop ecx pop ebx pop ebp ; restore stack pointer emms ; clear mmx registers ret ; we now return you to your regular programming _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/libavcodec/i386 - New directory
by Arpi of Ize 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/libavcodec/i386 In directory mplayer:/var/tmp.root/cvs-serv27239/i386 Log Message: Directory /cvsroot/mplayer/main/libavcodec/i386 added to the repository _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/libavcodec - New directory
by Arpi of Ize 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/libavcodec In directory mplayer:/var/tmp.root/cvs-serv27194/libavcodec Log Message: Directory /cvsroot/mplayer/main/libavcodec added to the repository _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main configure,1.98,1.99 Makefile,1.48,1.49
by Arpi of Ize 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main In directory mplayer:/var/tmp.root/cvs-serv26927 Modified Files: configure Makefile Log Message: ffmpeg support Index: configure =================================================================== RCS file: /cvsroot/mplayer/main/configure,v retrieving revision 1.98 retrieving revision 1.99 diff -u -r1.98 -r1.99 --- configure 4 Jul 2001 10:45:20 -0000 1.98 +++ configure 6 Jul 2001 03:22:14 -0000 1.99 @@ -377,6 +377,7 @@ _dshow=yes _fastmemcpy=yes _streaming=no +_libavcodec=no _x=1 _y=1 @@ -438,6 +439,11 @@ fi fi +if [ -d libavcodec ]; then + if [ -e libavcodec/Makefile ]; then + _libavcodec=yes + fi +fi if [ -c /dev/mga_vid ]; then _mga=yes @@ -1140,6 +1146,7 @@ echo "Checking for DeCSS support ... $_css" echo "Checking for PNG support ... $_png" echo "Checking for DirectShow ... $_dshow" +echo "Checking for libavcodec ... $_libavcodec" echo "Checking for fastmemcpy ... $_fastmemcpy" echo "Extra libs : $_extralibdir" echo "Extra headers : $_extraincdir" @@ -1192,6 +1199,16 @@ _dshow='#undef USE_DIRECTSHOW' fi +if [ $_libavcodec = yes ]; then + _lavclib='-Llibavcodec -lavcodec' + _lavcdep='libavcodec/libavcodec.a' + _libavcodec='#define USE_LIBAVCODEC' +else + _lavclib='' + _lavcdep='' + _libavcodec='#undef USE_LIBAVCODEC' +fi + if [ $_fastmemcpy = yes ]; then _fastmemcpy='#define USE_FASTMEMCPY' else @@ -1520,8 +1537,8 @@ CSS_LIB = $_csslib CSS_INC = $_cssinc SDL_INC = $_sdlcflags -DS_DEP = $_dshowdep -DS_LIB = $_dshowlib +DS_DEP = $_dshowdep $_lavcdep +DS_LIB = $_dshowlib $_lavclib ALSA_LIB = $_alsalib ESD_LIB = $_esdlib ARCH_LIBS = $_archlibs @@ -1608,6 +1625,9 @@ /* DirectShow support */ $_dshow + +/* ffmpeg's libavcodec support (requires libavcodec source) */ +$_libavcodec /* Use 3dnow/mmxext/sse/mmx optimized fast memcpy() [maybe buggy... signal 4]*/ $_fastmemcpy Index: Makefile =================================================================== RCS file: /cvsroot/mplayer/main/Makefile,v retrieving revision 1.48 retrieving revision 1.49 diff -u -r1.48 -r1.49 --- Makefile 26 Jun 2001 23:15:56 -0000 1.48 +++ Makefile 6 Jul 2001 03:22:14 -0000 1.49 @@ -40,6 +40,9 @@ loader/DirectShow/libDS_Filter.a: $(MAKE) -C loader/DirectShow +libavcodec/libavcodec.a: + $(MAKE) -C libavcodec + libmpeg2/libmpeg2.a: $(MAKE) -C libmpeg2 _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/DOCS codecs.conf,1.41,1.42
by Arpi of Ize 06 Jul '01

06 Jul '01

Update of /cvsroot/mplayer/main/DOCS In directory mplayer:/var/tmp.root/cvs-serv26762 Modified Files: codecs.conf Log Message: added Juanjo's H263+ patch, and ffmpeg native codecs Index: codecs.conf =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/codecs.conf,v retrieving revision 1.41 retrieving revision 1.42 diff -u -r1.41 -r1.42 --- codecs.conf 25 Jun 2001 14:06:14 -0000 1.41 +++ codecs.conf 6 Jul 2001 03:15:37 -0000 1.42 @@ -4,6 +4,38 @@ ; VIDEO CODECS ;============================================================================= +; =============== Win32/libmpeg2/ProjectMayo MPEG CODECS ===================== + +videocodec u263 + info "UB Video H.263+ Decoder" + comment "added by Juanjo" + status working + fourcc U263 + driver dshow + dll "ubv263d+.ax" + guid 0x00af1181, 0x6ebb, 0x11d4, 0x9d, 0x5a, 0x00, 0x50, 0x04, 0x79, 0x6c, 0xc0 +; out I420 flip +; out YVYU flip,noflip +; out YV12 flip +; out UYVY flip,noflip +; out YUY2 flip + out BGR24 flip + +videocodec divxds + info "DivX ;-) (MS MPEG-4 v3)" + comment "with postprocessing" + status working + fourcc MP43,mp43 div3 ; fourcc mapping to div3 + fourcc DIV5,div5 div3 + fourcc DIV6,div6 div4 + fourcc DIV3,div3,DIV4,div4 + driver dshow + dll "divx_c32.ax" + guid 0x82CCd3E0, 0xF71A, 0x11D0, 0x9f, 0xe5, 0x00, 0x60, 0x97, 0x78, 0xaa, 0xaa +; out YV12 + out YUY2 + out BGR32,BGR24,BGR16,BGR15 + videocodec mpeg12 info "MPEG 1 or 2" comment "with postprocessing" @@ -38,21 +70,6 @@ out YUY2 out BGR32,BGR24,BGR16,BGR15 -videocodec divxds - info "DivX ;-) (MS MPEG-4 v3)" - comment "with postprocessing" - status working - fourcc MP43,mp43 div3 ; fourcc mapping to div3 - fourcc DIV5,div5 div3 - fourcc DIV6,div6 div4 - fourcc DIV3,div3,DIV4,div4 - driver dshow - dll "divx_c32.ax" - guid 0x82CCd3E0, 0xF71A, 0x11D0, 0x9f, 0xe5, 0x00, 0x60, 0x97, 0x78, 0xaa, 0xaa -; out YV12 - out YUY2 - out BGR32,BGR24,BGR16,BGR15 - videocodec divx info "DivX ;-) (MS MPEG-4 v3)" status working @@ -76,6 +93,42 @@ driver odivx dll "opendivx decore" out YV12 + +; =============== FFMPEG MPEG CODECS ========================= + +videocodec ffdivx + info "FFmpeg DivX ;-) (MS MPEG-4 v3)" + status working + fourcc MP43,mp43 div3 ; M$ MPEG4 v3 (fourcc mapping to div3) + fourcc DIV5,div5 div3 ; DivX 3.20 + fourcc DIV6,div6 div4 ; -||- + fourcc DIV3,div3,DIV4,div4 + driver ffmpeg + dll msmpeg4 + out YV12 + +videocodec ffodivx + info "FFmpeg OpenDivX" + status working + fourcc DIVX,divx + fourcc DIV1,div1 divx + fourcc MP4S,mp4s ; ISO MPEG-4 Video V1 + format 0x4 + driver ffmpeg + dll opendivx + out YV12 + +videocodec ffmpeg12 + info "FFmpeg-1" + status working + format 0x10000001 + driver ffmpeg + dll "mpeg1video" + out YV12 + + +; ============== OTHER STUFF ======================= + videocodec indeo5ds info "Intel Indeo 5" _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/mp3lib decode_MMX.s,1.3,1.4 dct64_MMX.s,1.2,1.3 dct64_3dnow.s,1.3,1.4 dct64_k7.s,1.5,1.6
by Nick Kurshev 05 Jul '01

05 Jul '01

Update of /cvsroot/mplayer/main/mp3lib In directory mplayerhq:/var/tmp.root/cvs-serv9316/main/mp3lib Modified Files: decode_MMX.s dct64_MMX.s dct64_3dnow.s dct64_k7.s Log Message: Slight otimization Index: decode_MMX.s =================================================================== RCS file: /cvsroot/mplayer/main/mp3lib/decode_MMX.s,v retrieving revision 1.3 retrieving revision 1.4 diff -u -r1.3 -r1.4 --- decode_MMX.s 4 Jul 2001 09:47:56 -0000 1.3 +++ decode_MMX.s 5 Jul 2001 09:44:18 -0000 1.4 @@ -48,46 +48,48 @@ .text .globl synth_1to1_MMX_s - +/* + void synth_1to1_MMX_s(real *bandPtr, int channel, short *samples, + short *buffs, int *bo); +*/ synth_1to1_MMX_s: pushl %ebp pushl %edi pushl %esi pushl %ebx - movl 24(%esp),%ecx - movl 28(%esp),%edi + movl 24(%esp),%ecx + movl 28(%esp),%edi movl $15,%ebx - movl 36(%esp),%edx + movl 36(%esp),%edx leal (%edi,%ecx,2),%edi decl %ecx - movl 32(%esp),%esi - movl (%edx),%eax + movl 32(%esp),%esi + movl (%edx),%eax jecxz .L1 decl %eax - andl %ebx,%eax - leal 1088(%esi),%esi - movl %eax,(%edx) + andl %ebx,%eax + leal 1088(%esi),%esi + movl %eax,(%edx) .L1: - leal (%esi,%eax,2),%edx - movl %eax,%ebp - incl %eax - pushl 20(%esp) - andl %ebx,%eax - leal 544(%esi,%eax,2),%ecx - incl %ebx + leal (%esi,%eax,2),%edx + movl %eax,%ebp + incl %eax + pushl 20(%esp) + andl %ebx,%eax + leal 544(%esi,%eax,2),%ecx + incl %ebx testl $1, %eax - jnz .L2 + jnz .L2 xchgl %edx,%ecx incl %ebp - leal 544(%esi),%esi + leal 544(%esi),%esi .L2: emms pushl %edx pushl %ecx call *dct64_MMX_func - addl $12,%esp leal 1(%ebx), %ecx - subl %ebp,%ebx + subl %ebp,%ebx pushl %ecx leal decwins(%ebx,%ebx,1), %edx shrl $1, %ecx @@ -248,5 +250,3 @@ popl %edi popl %ebp ret - - Index: dct64_MMX.s =================================================================== RCS file: /cvsroot/mplayer/main/mp3lib/dct64_MMX.s,v retrieving revision 1.2 retrieving revision 1.3 diff -u -r1.2 -r1.3 --- dct64_MMX.s 4 Jul 2001 09:47:56 -0000 1.2 +++ dct64_MMX.s 5 Jul 2001 09:44:18 -0000 1.3 @@ -989,6 +989,4 @@ popl %edi popl %esi popl %ebx - ret - - + ret $12 Index: dct64_3dnow.s =================================================================== RCS file: /cvsroot/mplayer/main/mp3lib/dct64_3dnow.s,v retrieving revision 1.3 retrieving revision 1.4 diff -u -r1.3 -r1.4 --- dct64_3dnow.s 4 Jul 2001 09:47:56 -0000 1.3 +++ dct64_3dnow.s 5 Jul 2001 09:44:18 -0000 1.4 @@ -7,6 +7,11 @@ # Warning: Phases 7 & 8 are not tested # +.data + .align 8 +x_plus_minus_3dnow: .long 0x00000000, 0x80000000 +plus_1f: .float 1.0 + .text .align 16 @@ -501,23 +506,41 @@ /* Phase 6. This is the end of easy road. */ /* Code below is coded in scalar mode. Should be optimized */ - movd 32(%ecx), %mm0 - pfadd 36(%ecx), %mm0 - movd %mm0, 32(%edx) - - movd 32(%ecx), %mm0 - pfsub 36(%ecx), %mm0 - pfmul 120(%ebx),%mm0 - movd %mm0, 36(%edx) - - movd 44(%ecx), %mm0 - pfsub 40(%ecx), %mm0 - pfmul 120(%ebx),%mm0 - - movd %mm0, 44(%edx) - pfadd 40(%ecx), %mm0 - pfadd 44(%ecx), %mm0 - movd %mm0, 40(%edx) + movd plus_1f, %mm6 + punpckldq 120(%ebx), %mm6 /* mm6 = 1.0 | 120(%ebx)*/ + movq x_plus_minus_3dnow, %mm7 /* mm7 = +1 | -1 */ + + movq 32(%ecx), %mm0 + movq 64(%ecx), %mm2 + movq %mm0, %mm1 + movq %mm2, %mm3 + pxor %mm7, %mm1 + pxor %mm7, %mm3 + pfacc %mm1, %mm0 + pfacc %mm3, %mm2 + pfmul %mm6, %mm0 + pfmul %mm6, %mm2 + movq %mm0, 32(%edx) + movq %mm2, 64(%edx) + + movd 44(%ecx), %mm0 + movd 40(%ecx), %mm2 + movd 120(%ebx), %mm3 + punpckldq 76(%ecx), %mm0 + punpckldq 72(%ecx), %mm2 + punpckldq %mm3, %mm3 + movq %mm0, %mm4 + movq %mm2, %mm5 + pfsub %mm2, %mm0 + pfmul %mm3, %mm0 + movq %mm0, %mm1 + pfadd %mm5, %mm0 + pfadd %mm4, %mm0 + movq %mm0, %mm2 + punpckldq %mm1, %mm0 + punpckhdq %mm1, %mm2 + movq %mm0, 40(%edx) + movq %mm2, 72(%edx) movd 48(%ecx), %mm3 pfsub 52(%ecx), %mm3 @@ -534,38 +557,22 @@ pfadd 48(%ecx), %mm0 pfadd 52(%ecx), %mm0 - movd %mm0, 48(%edx) - pfadd %mm3, %mm1 - movd %mm1, 56(%edx) - movd %mm2, 60(%edx) - pfadd %mm3, %mm2 - movd %mm2, 52(%edx) + pfadd %mm3, %mm1 + punpckldq %mm2, %mm1 + pfadd %mm3, %mm2 + punpckldq %mm2, %mm0 + movq %mm1, 56(%edx) + movq %mm0, 48(%edx) /*---*/ - movd 64(%ecx), %mm0 - pfadd 68(%ecx), %mm0 - movd %mm0, 64(%edx) - - movd 64(%ecx), %mm0 - pfsub 68(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 68(%edx) - - movd 76(%ecx), %mm0 - pfsub 72(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 76(%edx) - pfadd 72(%ecx), %mm0 - pfadd 76(%ecx), %mm0 - movd %mm0, 72(%edx) - movd 92(%ecx), %mm0 - pfsub 88(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 92(%edx) - pfadd 92(%ecx), %mm0 - pfadd 88(%ecx), %mm0 - movq %mm0, %mm1 + movd 92(%ecx), %mm1 + pfsub 88(%ecx), %mm1 + pfmul 120(%ebx), %mm1 + movd %mm1, 92(%edx) + pfadd 92(%ecx), %mm1 + pfadd 88(%ecx), %mm1 + movq %mm1, %mm0 pfadd 80(%ecx), %mm0 pfadd 84(%ecx), %mm0 @@ -579,14 +586,12 @@ movd %mm0, 84(%edx) movd %mm1, 88(%edx) - movd 96(%ecx), %mm0 - pfadd 100(%ecx), %mm0 - movd %mm0, 96(%edx) - - movd 96(%ecx), %mm0 - pfsub 100(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 100(%edx) + movq 96(%ecx), %mm0 + movq %mm0, %mm1 + pxor %mm7, %mm1 + pfacc %mm1, %mm0 + pfmul %mm6, %mm0 + movq %mm0, 96(%edx) movd 108(%ecx), %mm0 pfsub 104(%ecx), %mm0 @@ -596,13 +601,13 @@ pfadd 108(%ecx), %mm0 movd %mm0, 104(%edx) - movd 124(%ecx), %mm0 - pfsub 120(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 124(%edx) - pfadd 120(%ecx), %mm0 - pfadd 124(%ecx), %mm0 - movq %mm0, %mm1 + movd 124(%ecx), %mm1 + pfsub 120(%ecx), %mm1 + pfmul 120(%ebx), %mm1 + movd %mm1, 124(%edx) + pfadd 120(%ecx), %mm1 + pfadd 124(%ecx), %mm1 + movq %mm1, %mm0 pfadd 112(%ecx), %mm0 pfadd 116(%ecx), %mm0 @@ -738,18 +743,18 @@ jmp .L_bye .L01: /* Phase 9*/ - movd (%ecx), %mm0 - pfadd 4(%ecx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 512(%esi) - movd (%ecx), %mm0 - pfsub 4(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, (%esi) + movq (%ecx), %mm0 + movq %mm0, %mm1 + pxor %mm7, %mm1 + pfacc %mm1, %mm0 + pfmul %mm6, %mm0 + pf2id %mm0, %mm0 + movd %mm0, %eax + movw %ax, 512(%esi) + psrlq $32, %mm0 + movd %mm0, %eax + movw %ax, (%esi) movd 12(%ecx), %mm0 pfsub 8(%ecx), %mm0 @@ -763,22 +768,24 @@ movd %mm0, %eax movw %ax, 256(%esi) - movd 16(%ecx), %mm0 - pfsub 20(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movq %mm0, %mm3 + movd 16(%ecx), %mm3 + pfsub 20(%ecx), %mm3 + pfmul 120(%ebx), %mm3 + movq %mm3, %mm2 - movd 28(%ecx), %mm0 - pfsub 24(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - pf2id %mm0, %mm7 + movd 28(%ecx), %mm2 + pfsub 24(%ecx), %mm2 + pfmul 120(%ebx), %mm2 + movq %mm2, %mm1 + + pf2id %mm2, %mm7 movd %mm7, %eax movw %ax, 384(%edi) - movq %mm0, %mm2 - pfadd 24(%ecx), %mm0 - pfadd 28(%ecx), %mm0 - movq %mm0, %mm1 + pfadd 24(%ecx), %mm1 + pfadd 28(%ecx), %mm1 + movq %mm1, %mm0 + pfadd 16(%ecx), %mm0 pfadd 20(%ecx), %mm0 pf2id %mm0, %mm0 @@ -793,7 +800,6 @@ movd %mm2, %eax movw %ax, 128(%edi) - /* Phase 10*/ movq 32(%edx), %mm0 @@ -813,32 +819,51 @@ movw %ax, 64(%edi) movw %cx, 192(%edi) - movd 40(%edx), %mm0 - pfadd 56(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 192(%esi) - - movd 56(%edx), %mm0 - pfadd 36(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 64(%esi) - + movd 40(%edx), %mm3 + movd 56(%edx), %mm4 movd 60(%edx), %mm0 - pf2id %mm0, %mm7 - movd %mm7, %eax + movd 44(%edx), %mm2 + movd 120(%edx), %mm5 + punpckldq %mm4, %mm3 + punpckldq 124(%edx), %mm0 + pfadd 100(%edx), %mm5 + punpckldq 36(%edx), %mm4 + punpckldq 92(%edx), %mm2 + movq %mm5, %mm6 + pfadd %mm4, %mm3 + pf2id %mm0, %mm1 + pf2id %mm3, %mm3 + pfadd 88(%edx), %mm5 + movd %mm1, %eax + movd %mm3, %ecx movw %ax, 448(%edi) - pfadd 44(%edx), %mm0 + movw %cx, 192(%esi) + pf2id %mm5, %mm5 + psrlq $32, %mm1 + psrlq $32, %mm3 + movd %mm5, %ebx + movd %mm1, %eax + movd %mm3, %ecx + movw %bx, 96(%esi) + movw %ax, 480(%edi) + movw %cx, 64(%esi) + pfadd %mm2, %mm0 pf2id %mm0, %mm0 movd %mm0, %eax + pfadd 68(%edx), %mm6 movw %ax, 320(%edi) + psrlq $32, %mm0 + pf2id %mm6, %mm6 + movd %mm0, %eax + movd %mm6, %ebx + movw %ax, 416(%edi) + movw %bx, 32(%esi) movq 96(%edx), %mm0 movq 112(%edx), %mm2 movq 104(%edx), %mm4 - pfadd 112(%edx), %mm0 - pfadd 104(%edx), %mm2 + pfadd %mm2, %mm0 + pfadd %mm4, %mm2 pfadd 120(%edx), %mm4 movq %mm0, %mm1 movq %mm2, %mm3 @@ -846,20 +871,20 @@ pfadd 64(%edx), %mm0 pfadd 80(%edx), %mm2 pfadd 72(%edx), %mm4 - pf2id %mm0, %mm7 - pf2id %mm2, %mm6 + pf2id %mm0, %mm0 + pf2id %mm2, %mm2 pf2id %mm4, %mm4 - movd %mm7, %eax - movd %mm6, %ecx + movd %mm0, %eax + movd %mm2, %ecx movd %mm4, %ebx movw %ax, 480(%esi) movw %cx, 352(%esi) movw %bx, 224(%esi) - psrlq $32, %mm7 - psrlq $32, %mm6 + psrlq $32, %mm0 + psrlq $32, %mm2 psrlq $32, %mm4 - movd %mm7, %eax - movd %mm6, %ecx + movd %mm0, %eax + movd %mm2, %ecx movd %mm4, %ebx movw %ax, 32(%edi) movw %cx, 160(%edi) @@ -886,27 +911,6 @@ movw %cx, 224(%edi) movw %bx, 352(%edi) - movd 120(%edx), %mm0 - pfadd 100(%edx), %mm0 - movq %mm0, %mm1 - pfadd 88(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 96(%esi) - pfadd 68(%edx), %mm1 - pf2id %mm1, %mm1 - movd %mm1, %eax - movw %ax, 32(%esi) - - movq 124(%edx), %mm0 - pf2id %mm0, %mm1 - movd %mm1, %eax - movw %ax, 480(%edi) - pfadd 92(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 416(%edi) - movsw .L_bye: @@ -915,6 +919,6 @@ popl %edi popl %esi popl %ebx - ret - + ret $12 + Index: dct64_k7.s =================================================================== RCS file: /cvsroot/mplayer/main/mp3lib/dct64_k7.s,v retrieving revision 1.5 retrieving revision 1.6 diff -u -r1.5 -r1.6 --- dct64_k7.s 4 Jul 2001 09:47:56 -0000 1.5 +++ dct64_k7.s 5 Jul 2001 09:44:18 -0000 1.6 @@ -7,6 +7,11 @@ # Warning: Phases 7 & 8 are not tested # +.data + .align 8 +x_plus_minus_3dnow: .long 0x00000000, 0x80000000 +plus_1f: .float 1.0 + .text .align 16 @@ -201,6 +206,9 @@ /* Phase 4*/ + movq 96(%ebx), %mm2 + movq 104(%ebx), %mm6 + movq (%ecx), %mm0 movq 8(%ecx), %mm4 movq %mm0, %mm3 @@ -215,8 +223,8 @@ movq %mm4, 8(%edx) pfsub %mm1, %mm3 pfsub %mm5, %mm7 - pfmul 96(%ebx), %mm3 - pfmul 104(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm6, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 24(%edx) @@ -236,8 +244,8 @@ movq %mm4, 40(%edx) pfsubr %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 96(%ebx), %mm3 - pfmul 104(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm6, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 56(%edx) @@ -257,8 +265,8 @@ movq %mm4, 72(%edx) pfsub %mm1, %mm3 pfsub %mm5, %mm7 - pfmul 96(%ebx), %mm3 - pfmul 104(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm6, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 88(%edx) @@ -278,8 +286,8 @@ movq %mm4, 104(%edx) pfsubr %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 96(%ebx), %mm3 - pfmul 104(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm6, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 120(%edx) @@ -287,6 +295,8 @@ /* Phase 5 */ + movq 112(%ebx), %mm2 + movq (%edx), %mm0 movq 16(%edx), %mm4 movq %mm0, %mm3 @@ -301,8 +311,8 @@ movq %mm4, 16(%ecx) pfsub %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 112(%ebx), %mm3 - pfmul 112(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm2, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 8(%ecx) @@ -322,8 +332,8 @@ movq %mm4, 48(%ecx) pfsub %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 112(%ebx), %mm3 - pfmul 112(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm2, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 40(%ecx) @@ -343,8 +353,8 @@ movq %mm4, 80(%ecx) pfsub %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 112(%ebx), %mm3 - pfmul 112(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm2, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 72(%ecx) @@ -364,33 +374,52 @@ movq %mm4, 112(%ecx) pfsub %mm1, %mm3 pfsubr %mm5, %mm7 - pfmul 112(%ebx), %mm3 - pfmul 112(%ebx), %mm7 + pfmul %mm2, %mm3 + pfmul %mm2, %mm7 pswapd %mm3, %mm3 pswapd %mm7, %mm7 movq %mm3, 104(%ecx) movq %mm7, 120(%ecx) + /* Phase 6. This is the end of easy road. */ /* Code below is coded in scalar mode. Should be optimized */ - movd 32(%ecx), %mm0 - pfadd 36(%ecx), %mm0 - movd %mm0, 32(%edx) - - movd 32(%ecx), %mm0 - pfsub 36(%ecx), %mm0 - pfmul 120(%ebx),%mm0 - movd %mm0, 36(%edx) - - movd 44(%ecx), %mm0 - pfsub 40(%ecx), %mm0 - pfmul 120(%ebx),%mm0 - - movd %mm0, 44(%edx) - pfadd 40(%ecx), %mm0 - pfadd 44(%ecx), %mm0 - movd %mm0, 40(%edx) + movd plus_1f, %mm6 + punpckldq 120(%ebx), %mm6 /* mm6 = 1.0 | 120(%ebx)*/ + movq x_plus_minus_3dnow, %mm7 /* mm7 = +1 | -1 */ + + movq 32(%ecx), %mm0 + movq 64(%ecx), %mm2 + movq %mm0, %mm1 + movq %mm2, %mm3 + pxor %mm7, %mm1 + pxor %mm7, %mm3 + pfacc %mm1, %mm0 + pfacc %mm3, %mm2 + pfmul %mm6, %mm0 + pfmul %mm6, %mm2 + movq %mm0, 32(%edx) + movq %mm2, 64(%edx) + + movd 44(%ecx), %mm0 + movd 40(%ecx), %mm2 + movd 120(%ebx), %mm3 + punpckldq 76(%ecx), %mm0 + punpckldq 72(%ecx), %mm2 + punpckldq %mm3, %mm3 + movq %mm0, %mm4 + movq %mm2, %mm5 + pfsub %mm2, %mm0 + pfmul %mm3, %mm0 + movq %mm0, %mm1 + pfadd %mm5, %mm0 + pfadd %mm4, %mm0 + movq %mm0, %mm2 + punpckldq %mm1, %mm0 + punpckhdq %mm1, %mm2 + movq %mm0, 40(%edx) + movq %mm2, 72(%edx) movd 48(%ecx), %mm3 pfsub 52(%ecx), %mm3 @@ -407,38 +436,22 @@ pfadd 48(%ecx), %mm0 pfadd 52(%ecx), %mm0 - movd %mm0, 48(%edx) - pfadd %mm3, %mm1 - movd %mm1, 56(%edx) - movd %mm2, 60(%edx) - pfadd %mm3, %mm2 - movd %mm2, 52(%edx) + pfadd %mm3, %mm1 + punpckldq %mm2, %mm1 + pfadd %mm3, %mm2 + punpckldq %mm2, %mm0 + movq %mm1, 56(%edx) + movq %mm0, 48(%edx) /*---*/ - movd 64(%ecx), %mm0 - pfadd 68(%ecx), %mm0 - movd %mm0, 64(%edx) - movd 64(%ecx), %mm0 - pfsub 68(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 68(%edx) - - movd 76(%ecx), %mm0 - pfsub 72(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 76(%edx) - pfadd 72(%ecx), %mm0 - pfadd 76(%ecx), %mm0 - movd %mm0, 72(%edx) - - movd 92(%ecx), %mm0 - pfsub 88(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 92(%edx) - pfadd 92(%ecx), %mm0 - pfadd 88(%ecx), %mm0 - movq %mm0, %mm1 + movd 92(%ecx), %mm1 + pfsub 88(%ecx), %mm1 + pfmul 120(%ebx), %mm1 + movd %mm1, 92(%edx) + pfadd 92(%ecx), %mm1 + pfadd 88(%ecx), %mm1 + movq %mm1, %mm0 pfadd 80(%ecx), %mm0 pfadd 84(%ecx), %mm0 @@ -452,14 +465,12 @@ movd %mm0, 84(%edx) movd %mm1, 88(%edx) - movd 96(%ecx), %mm0 - pfadd 100(%ecx), %mm0 - movd %mm0, 96(%edx) - - movd 96(%ecx), %mm0 - pfsub 100(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 100(%edx) + movq 96(%ecx), %mm0 + movq %mm0, %mm1 + pxor %mm7, %mm1 + pfacc %mm1, %mm0 + pfmul %mm6, %mm0 + movq %mm0, 96(%edx) movd 108(%ecx), %mm0 pfsub 104(%ecx), %mm0 @@ -469,13 +480,13 @@ pfadd 108(%ecx), %mm0 movd %mm0, 104(%edx) - movd 124(%ecx), %mm0 - pfsub 120(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movd %mm0, 124(%edx) - pfadd 120(%ecx), %mm0 - pfadd 124(%ecx), %mm0 - movq %mm0, %mm1 + movd 124(%ecx), %mm1 + pfsub 120(%ecx), %mm1 + pfmul 120(%ebx), %mm1 + movd %mm1, 124(%edx) + pfadd 120(%ecx), %mm1 + pfadd 124(%ecx), %mm1 + movq %mm1, %mm0 pfadd 112(%ecx), %mm0 pfadd 116(%ecx), %mm0 @@ -611,18 +622,18 @@ jmp .L_bye .L01: /* Phase 9*/ - movd (%ecx), %mm0 - pfadd 4(%ecx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 512(%esi) - movd (%ecx), %mm0 - pfsub 4(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, (%esi) + movq (%ecx), %mm0 + movq %mm0, %mm1 + pxor %mm7, %mm1 + pfacc %mm1, %mm0 + pfmul %mm6, %mm0 + pf2id %mm0, %mm0 + movd %mm0, %eax + movw %ax, 512(%esi) + psrlq $32, %mm0 + movd %mm0, %eax + movw %ax, (%esi) movd 12(%ecx), %mm0 pfsub 8(%ecx), %mm0 @@ -636,22 +647,24 @@ movd %mm0, %eax movw %ax, 256(%esi) - movd 16(%ecx), %mm0 - pfsub 20(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - movq %mm0, %mm3 + movd 16(%ecx), %mm3 + pfsub 20(%ecx), %mm3 + pfmul 120(%ebx), %mm3 + movq %mm3, %mm2 - movd 28(%ecx), %mm0 - pfsub 24(%ecx), %mm0 - pfmul 120(%ebx), %mm0 - pf2id %mm0, %mm7 + movd 28(%ecx), %mm2 + pfsub 24(%ecx), %mm2 + pfmul 120(%ebx), %mm2 + movq %mm2, %mm1 + + pf2id %mm2, %mm7 movd %mm7, %eax movw %ax, 384(%edi) - movq %mm0, %mm2 - pfadd 24(%ecx), %mm0 - pfadd 28(%ecx), %mm0 - movq %mm0, %mm1 + pfadd 24(%ecx), %mm1 + pfadd 28(%ecx), %mm1 + movq %mm1, %mm0 + pfadd 16(%ecx), %mm0 pfadd 20(%ecx), %mm0 pf2id %mm0, %mm0 @@ -666,7 +679,6 @@ movd %mm2, %eax movw %ax, 128(%edi) - /* Phase 10*/ movq 32(%edx), %mm0 @@ -686,32 +698,51 @@ movw %ax, 64(%edi) movw %cx, 192(%edi) - movd 40(%edx), %mm0 - pfadd 56(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 192(%esi) - - movd 56(%edx), %mm0 - pfadd 36(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 64(%esi) - + movd 40(%edx), %mm3 + movd 56(%edx), %mm4 movd 60(%edx), %mm0 - pf2id %mm0, %mm7 - movd %mm7, %eax + movd 44(%edx), %mm2 + movd 120(%edx), %mm5 + punpckldq %mm4, %mm3 + punpckldq 124(%edx), %mm0 + pfadd 100(%edx), %mm5 + punpckldq 36(%edx), %mm4 + punpckldq 92(%edx), %mm2 + movq %mm5, %mm6 + pfadd %mm4, %mm3 + pf2id %mm0, %mm1 + pf2id %mm3, %mm3 + pfadd 88(%edx), %mm5 + movd %mm1, %eax + movd %mm3, %ecx movw %ax, 448(%edi) - pfadd 44(%edx), %mm0 + movw %cx, 192(%esi) + pf2id %mm5, %mm5 + psrlq $32, %mm1 + psrlq $32, %mm3 + movd %mm5, %ebx + movd %mm1, %eax + movd %mm3, %ecx + movw %bx, 96(%esi) + movw %ax, 480(%edi) + movw %cx, 64(%esi) + pfadd %mm2, %mm0 pf2id %mm0, %mm0 movd %mm0, %eax + pfadd 68(%edx), %mm6 movw %ax, 320(%edi) + psrlq $32, %mm0 + pf2id %mm6, %mm6 + movd %mm0, %eax + movd %mm6, %ebx + movw %ax, 416(%edi) + movw %bx, 32(%esi) movq 96(%edx), %mm0 movq 112(%edx), %mm2 movq 104(%edx), %mm4 - pfadd 112(%edx), %mm0 - pfadd 104(%edx), %mm2 + pfadd %mm2, %mm0 + pfadd %mm4, %mm2 pfadd 120(%edx), %mm4 movq %mm0, %mm1 movq %mm2, %mm3 @@ -719,20 +750,20 @@ pfadd 64(%edx), %mm0 pfadd 80(%edx), %mm2 pfadd 72(%edx), %mm4 - pf2id %mm0, %mm7 - pf2id %mm2, %mm6 + pf2id %mm0, %mm0 + pf2id %mm2, %mm2 pf2id %mm4, %mm4 - movd %mm7, %eax - movd %mm6, %ecx + movd %mm0, %eax + movd %mm2, %ecx movd %mm4, %ebx movw %ax, 480(%esi) movw %cx, 352(%esi) movw %bx, 224(%esi) - psrlq $32, %mm7 - psrlq $32, %mm6 + psrlq $32, %mm0 + psrlq $32, %mm2 psrlq $32, %mm4 - movd %mm7, %eax - movd %mm6, %ecx + movd %mm0, %eax + movd %mm2, %ecx movd %mm4, %ebx movw %ax, 32(%edi) movw %cx, 160(%edi) @@ -759,27 +790,6 @@ movw %cx, 224(%edi) movw %bx, 352(%edi) - movd 120(%edx), %mm0 - pfadd 100(%edx), %mm0 - movq %mm0, %mm1 - pfadd 88(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 96(%esi) - pfadd 68(%edx), %mm1 - pf2id %mm1, %mm1 - movd %mm1, %eax - movw %ax, 32(%esi) - - movq 124(%edx), %mm0 - pf2id %mm0, %mm1 - movd %mm1, %eax - movw %ax, 480(%edi) - pfadd 92(%edx), %mm0 - pf2id %mm0, %mm0 - movd %mm0, %eax - movw %ax, 416(%edi) - movsw .L_bye: @@ -788,6 +798,4 @@ popl %edi popl %esi popl %ebx - ret - - + ret $12 _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/mp3lib sr1.c,1.10,1.11
by Nick Kurshev 04 Jul '01

04 Jul '01

Update of /cvsroot/mplayer/main/mp3lib In directory mplayerhq:/var/tmp.root/cvs-serv22195/main/mp3lib Modified Files: sr1.c Log Message: One flaw is fixed Index: sr1.c =================================================================== RCS file: /cvsroot/mplayer/main/mp3lib/sr1.c,v retrieving revision 1.10 retrieving revision 1.11 diff -u -r1.10 -r1.11 --- sr1.c 3 Jul 2001 07:50:52 -0000 1.10 +++ sr1.c 4 Jul 2001 20:54:09 -0000 1.11 @@ -394,9 +394,13 @@ _CpuID = _i586 = _3dnow = _isse = _has_mmx = 0; printf( "mp3lib: Using generic decore.\n"); #endif +#ifdef HAVE_MMX /* Use it for any MMX cpu */ - if(_has_mmx) make_decode_tables_MMX(outscale); - else make_decode_tables(outscale); + if(_has_mmx) + make_decode_tables_MMX(outscale); + else +#endif + make_decode_tables(outscale); #ifdef USE_FAKE_MONO if (fakemono == 1) fr.synth=synth_1to1_l; _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0

CVS: main/DOCS MPlayer-FAQ,1.55,1.56 example.conf,1.25,1.26 mplayer.1,1.61,1.62
by Winner of tha face compo 04 Jul '01

04 Jul '01

Update of /cvsroot/mplayer/main/DOCS In directory mplayerhq:/var/tmp.root/cvs-serv20731/DOCS Modified Files: MPlayer-FAQ example.conf mplayer.1 Log Message: ICCCCMSFNSFN Index: MPlayer-FAQ =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/MPlayer-FAQ,v retrieving revision 1.55 retrieving revision 1.56 diff -u -r1.55 -r1.56 --- MPlayer-FAQ 1 Jul 2001 21:49:21 -0000 1.55 +++ MPlayer-FAQ 4 Jul 2001 20:18:24 -0000 1.56 @@ -206,14 +206,7 @@ Q: I have problems with <your window manager> and fullscreen xv/xmga/sdl/x11 modes.. -A: Yes, it's a bug. But not only our bug. We couldn't yet develop any - windowmanager-independent method to make window really fullscreen. - There was some version, but there are always users with problems. - I think the only solution is adding a new switch, for example -fsmode - to select which method should be used. - Unfortunately most common window managers aren't compatible with standards - or Motif wm. So standard ways to enable fullscreen doesn't work with them. - And non-standard modes not work for standard-compliant wm's. +A1: Use the -fsmode switch. See example.conf or manpage. Q: But it works with avifile ! A: So what? Index: example.conf =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/example.conf,v retrieving revision 1.25 retrieving revision 1.26 diff -u -r1.25 -r1.26 --- example.conf 23 Jun 2001 22:42:38 -0000 1.25 +++ example.conf 4 Jul 2001 20:18:24 -0000 1.26 @@ -17,6 +17,16 @@ fs=yes # Enlarges movie window to your desktop's size. # Used by drivers: all +# fsmode=0 # This option workarounds some problems when using + # specific windowmanagers and fullscreen mode. If you + # experience fullscreen problems, try changing this + # value between 0 and 7. Hint: either 0 or 1 should + # be sufficient. + # 0 = new method (since 0.18pre3) + # 1 = ICCCWM patch (for KDE2/icewm) + # 2 = old method (0.17a) + # 3 = ICCCWM patch + Motif method + vm=no # Tries to change to a different videomode # Used by drivers: dga2, x11, sdl Index: mplayer.1 =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/mplayer.1,v retrieving revision 1.61 retrieving revision 1.62 diff -u -r1.61 -r1.62 --- mplayer.1 4 Jul 2001 16:56:31 -0000 1.61 +++ mplayer.1 4 Jul 2001 20:18:24 -0000 1.62 @@ -288,6 +288,7 @@ -fsmode 0 new method (since 0.18pre3) -fsmode 1 ICCCWM patch (for KDE2/icewm) -fsmode 2 old method (0.17a) + -fsmode 3 ICCCWM patch + Motif method .IP .SH "ALPHA/BETA CODE" .TP _______________________________________________ Mplayer-cvslog mailing list Mplayer-cvslog(a)lists.sourceforge.net http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog

1 0