vdec_idctmmx.s

vlc stand 0.1.99 ist sehr einfach

/*****************************************************************************
 * vdec_idctmmx.S : MMX IDCT implementation
 *****************************************************************************
 * Copyright (C) 1999, 2000 VideoLAN
 *
 * Authors:
 *
 * 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, USA.
 *****************************************************************************/

/*
 * the input data is tranposed and each 16 bit element in the 8x8 matrix
 * is left aligned:
 * for example in 11...1110000 format
 * If the iDCT is of I macroblock then 0.5 needs to be added to the;DC Component
 * (element[0][0] of the matrix)
 */

.data
        .align 16
        .type    preSC,@object
preSC:  .short  16384,22725,21407,19266,16384,12873,8867,4520
        .short  22725,31521,29692,26722,22725,17855,12299,6270
        .short  21407,29692,27969,25172,21407,16819,11585,5906
        .short  19266,26722,25172,22654,19266,15137,10426,5315
        .short  16384,22725,21407,19266,16384,12873,8867,4520
        .short  12873,17855,16819,15137,25746,20228,13933,7103
        .short  17734,24598,23170,20853,17734,13933,9597,4892
        .short  18081,25080,23624,21261,18081,14206,9785,4988
        .size    preSC,128
        .align 8
        .type   x0005000200010001,@object
        .size   x0005000200010001,8
x0005000200010001:
        .long   0x00010001,0x00050002
        .align 8
        .type   x0040000000000000,@object
        .size   x0040000000000000,8
x0040000000000000:
        .long   0, 0x00400000
        .align 8
        .type   x5a825a825a825a82,@object
        .size   x5a825a825a825a82,8
x5a825a825a825a82:
        .long   0x5a825a82, 0x5a825a82
        .align 8
        .type   x539f539f539f539f,@object
        .size   x539f539f539f539f,8
x539f539f539f539f:
        .long   0x539f539f,0x539f539f
        .align 8
        .type   x4546454645464546,@object
        .size   x4546454645464546,8
x4546454645464546:
        .long   0x45464546,0x45464546
        .align 8
        .type   x61f861f861f861f8,@object
        .size   x61f861f861f861f8,8
x61f861f861f861f8:
        .long   0x61f861f8,0x61f861f8
        .align 8
        .type    scratch1,@object
        .size    scratch1,8
scratch1:
        .long 0,0
        .align 8
        .type    scratch3,@object
        .size    scratch3,8
scratch3:
        .long 0,0
        .align 8
        .type    scratch5,@object
        .size    scratch5,8
scratch5:
        .long 0,0
        .align 8
        .type    scratch7,@object
        .size    scratch7,8
scratch7:
        .long 0,0
        .type    x0,@object
        .size    x0,8
x0:
        .long 0,0
        .align 8
.text
        .align 4
.globl vdec_IDCT
        .type    vdec_IDCT,@function
vdec_IDCT:
        pushl %ebp
        movl %esp,%ebp
        pushl %ebx
        pushl %ecx
        pushl %edx
        pushl %esi
        pushl %edi
        leal preSC, %ecx
        movl 12(%ebp),%esi
        movq (%esi), %mm0
        movq 8(%esi), %mm1
        movq 16(%esi), %mm2
        movq 24(%esi), %mm3
        movq 32(%esi), %mm4
        movq 40(%esi), %mm5
        movq 48(%esi), %mm6
        movq 56(%esi), %mm7
        psllw $4, %mm0
        psllw $4, %mm1
        psllw $4, %mm2
        psllw $4, %mm3
        psllw $4, %mm4
        psllw $4, %mm5
        psllw $4, %mm6
        psllw $4, %mm7
        movq %mm0,  (%esi)
        movq %mm1, 8(%esi)
        movq %mm2,16(%esi)
        movq %mm3,24(%esi)
        movq %mm4,32(%esi)
        movq %mm5,40(%esi)
        movq %mm6,48(%esi)
        movq %mm7,56(%esi)
        movq 64(%esi), %mm0
        movq 72(%esi), %mm1
        movq 80(%esi), %mm2
        movq 88(%esi), %mm3
        movq 96(%esi), %mm4
        movq 104(%esi), %mm5
        movq 112(%esi), %mm6
        movq 120(%esi), %mm7
        psllw $4, %mm0
        psllw $4, %mm1
        psllw $4, %mm2
        psllw $4, %mm3
        psllw $4, %mm4
        psllw $4, %mm5
        psllw $4, %mm6
        psllw $4, %mm7
        movq %mm0,64(%esi)
        movq %mm1,72(%esi)
        movq %mm2,80(%esi)
        movq %mm3,88(%esi)
        movq %mm4,96(%esi)
        movq %mm5,104(%esi)
        movq %mm6,112(%esi)
        movq %mm7,120(%esi)
/* column 0: even part
 * use V4, V12, V0, V8 to produce V22..V25
 */
        movq 8*12(%ecx), %mm0   /* maybe the first mul can be done together */
                                /* with the dequantization in iHuff module */
        pmulhw 8*12(%esi), %mm0         /* V12 */
        movq 8*4(%ecx), %mm1
        pmulhw 8*4(%esi), %mm1		/* V4 */
        movq (%ecx), %mm3
        psraw $1, %mm0			/* t64=t66 */
        pmulhw (%esi), %mm3		/* V0 */
        movq 8*8(%ecx), %mm5		/* duplicate V4 */
        movq %mm1, %mm2			/* added 11/1/96 */
        pmulhw 8*8(%esi),%mm5		/* V8 */
        psubsw %mm0, %mm1		/* V16 */
        pmulhw x5a825a825a825a82, %mm1	/* 23170 ->V18 */
        paddsw %mm0, %mm2		/* V17 */
        movq %mm2, %mm0			/* duplicate V17 */
        psraw $1, %mm2			/* t75=t82 */
        psraw $2, %mm0			/* t72 */
        movq %mm3, %mm4			/* duplicate V0 */
        paddsw %mm5, %mm3		/* V19 */
        psubsw %mm5, %mm4		/* V20 ;mm5 free */
/* moved from the block below */
        movq 8*10(%ecx), %mm7
        psraw $1, %mm3			/* t74=t81 */
        movq %mm3, %mm6			/* duplicate t74=t81 */
        psraw $2, %mm4			/* t77=t79 */
        psubsw %mm0, %mm1		/* V21 ; mm0 free */
        paddsw %mm2, %mm3		/* V22 */
        movq %mm1, %mm5			/* duplicate V21 */
        paddsw %mm4, %mm1		/* V23 */
        movq %mm3, 8*4(%esi)		/* V22 */
        psubsw %mm5, %mm4		/* V24; mm5 free */
        movq %mm1, 8*12(%esi)		/* V23 */
        psubsw %mm2, %mm6		/* V25; mm2 free */
        movq %mm4, (%esi)		/* V24 */
/* keep mm6 alive all along the next block */
        /* movq %mm6, 8*8(%esi) 	V25 */
/* column 0: odd part
 * use V2, V6, V10, V14 to produce V31, V39, V40, V41
 */
/* moved above: movq 8*10(%ecx), %mm7 */

        pmulhw 8*10(%esi), %mm7		/* V10 */
        movq 8*6(%ecx), %mm0
        pmulhw 8*6(%esi), %mm0		/* V6 */
        movq 8*2(%ecx), %mm5
        movq %mm7, %mm3			/* duplicate V10 */
        pmulhw 8*2(%esi), %mm5		/* V2 */
        movq 8*14(%ecx), %mm4
        psubsw %mm0, %mm7		/* V26 */
        pmulhw 8*14(%esi), %mm4		/* V14 */
        paddsw %mm0, %mm3		/* V29 ; free mm0 */
        movq %mm7, %mm1			/* duplicate V26 */
        psraw $1, %mm3			/* t91=t94 */
        pmulhw x539f539f539f539f,%mm7	/* V33 */
        psraw $1, %mm1			/* t96 */
        movq %mm5, %mm0			/* duplicate V2 */
        psraw $2, %mm4			/* t85=t87 */
        paddsw %mm4,%mm5		/* V27 */
        psubsw %mm4, %mm0		/* V28 ; free mm4 */
        movq %mm0, %mm2			/* duplicate V28 */
        psraw $1, %mm5			/* t90=t93 */
        pmulhw x4546454645464546,%mm0	/* V35 */
        psraw $1, %mm2			/* t97 */
        movq %mm5, %mm4			/* duplicate t90=t93 */
        psubsw %mm2, %mm1		/* V32 ; free mm2 */
        pmulhw x61f861f861f861f8,%mm1	/* V36 */
        psllw $1, %mm7			/* t107 */
        paddsw %mm3, %mm5		/* V31 */
        psubsw %mm3, %mm4		/* V30 ; free mm3 */
        pmulhw x5a825a825a825a82,%mm4	/* V34 */
        nop
        psubsw %mm1, %mm0		/* V38 */
        psubsw %mm7, %mm1		/* V37 ; free mm7 */
        psllw $1, %mm1			/* t114 */
/* move from the next block */
        movq %mm6, %mm3			/* duplicate V25 */
/* move from the next block */
        movq 8*4(%esi), %mm7		/* V22 */
        psllw $1, %mm0			/* t110 */
        psubsw %mm5, %mm0		/* V39 (mm5 needed for next block) */
        psllw $2, %mm4			/* t112 */
/* moved from the next block */
        movq 8*12(%esi), %mm2		/* V23 */
        psubsw %mm0, %mm4		/* V40 */
        paddsw %mm4, %mm1		/* V41; free mm0 */
/* moved from the next block */
        psllw $1, %mm2			/* t117=t125 */
/* column 0: output butterfly */
/* moved above:
 * movq %mm6, %mm3			duplicate V25
 * movq 8*4(%esi), %mm7			V22
 * movq 8*12(%esi), %mm2		V23
 * psllw $1, %mm2			t117=t125
 */
        psubsw %mm1, %mm6		/* tm6 */
        paddsw %mm1, %mm3		/* tm8; free mm1 */
        movq %mm7, %mm1			/* duplicate V22 */
        paddsw %mm5, %mm7		/* tm0 */
        movq %mm3, 8*8(%esi)		/* tm8; free mm3 */
        psubsw %mm5, %mm1		/* tm14; free mm5 */
        movq %mm6, 8*6(%esi)		/* tm6; free mm6 */
        movq %mm2, %mm3			/* duplicate t117=t125 */
        movq (%esi), %mm6		/* V24 */
        paddsw %mm0, %mm2		/* tm2 */
        movq %mm7, (%esi)		/* tm0; free mm7 */
        psubsw %mm0, %mm3		/* tm12; free mm0 */
        movq %mm1, 8*14(%esi)		/* tm14; free mm1 */
        psllw $1, %mm6			/* t119=t123 */
        movq %mm2, 8*2(%esi)		/* tm2; free mm2 */
        movq %mm6, %mm0			/* duplicate t119=t123 */
        movq %mm3, 8*12(%esi)		/* tm12; free mm3 */
        paddsw %mm4, %mm6		/* tm4 */
/* moved from next block */
        movq 8*5(%ecx), %mm1