fdct_mmx.asm

视频压缩标准MPEG4的视频参考代码xvid9.1

;/*****************************************************************************
; *
; *  XVID MPEG-4 VIDEO CODEC
; *  mmx version - fast discrete cosine transformation 
; *
; *  Initial version provided by Intel at AppNote AP-922       
; *  Copyright (C) 1999 Intel Corporation,                                     
; * 
; *  completed and corrected                        
; *  Copyright (C) 2000 - Royce Shih-Wea Liao <[e-mail]>,               
; * 
; *  ported to NASM and some minor changes                                      
; *  Copyright (C) 2001 - Michael Militzer <[e-mail]>                  
; *
; *  This file is part of XviD, a free MPEG-4 video encoder/decoder
; *
; *  XviD 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
; *
; *  Under section 8 of the GNU General Public License, the copyright
; *  holders of XVID explicitly forbid distribution in the following
; *  countries:
; *
; *    - Japan
; *    - United States of America
; *
; *  Linking XviD statically or dynamically with other modules is making a
; *  combined work based on XviD.  Thus, the terms and conditions of the
; *  GNU General Public License cover the whole combination.
; *
; *  As a special exception, the copyright holders of XviD give you
; *  permission to link XviD with independent modules that communicate with
; *  XviD solely through the VFW1.1 and DShow interfaces, regardless of the
; *  license terms of these independent modules, and to copy and distribute
; *  the resulting combined work under terms of your choice, provided that
; *  every copy of the combined work is accompanied by a complete copy of
; *  the source code of XviD (the version of XviD used to produce the
; *  combined work), being distributed under the terms of the GNU General
; *  Public License plus this exception.  An independent module is a module
; *  which is not derived from or based on XviD.
; *
; *  Note that people who make modified versions of XviD are not obligated
; *  to grant this special exception for their modified versions; it is
; *  their choice whether to do so.  The GNU General Public License gives
; *  permission to release a modified version without this exception; this
; *  exception also makes it possible to release a modified version which
; *  carries forward this exception.
; *
; * $Id: fdct_mmx.asm,v 1.6 2002/11/16 23:51:58 edgomez Exp $
; *
; *************************************************************************/

BITS 32

%macro cglobal 1 
	%ifdef PREFIX
		global _%1 
		%define %1 _%1
	%else
		global %1
	%endif
%endmacro

%define INP eax
%define TABLE ebx
%define TABLEF ebx
%define OUT ecx
%define round_frw_row edx

%define INP_1 eax + 16
%define INP_2 eax + 32
%define INP_3 eax + 48
%define INP_4 eax + 64
%define INP_5 eax + 80
%define INP_6 eax + 96
%define INP_7 eax + 112

%define OUT_1 ecx + 16
%define OUT_2 ecx + 32
%define OUT_3 ecx + 48
%define OUT_4 ecx + 64
%define OUT_5 ecx + 80
%define OUT_6 ecx + 96
%define OUT_7 ecx + 112
%define OUT_8 ecx + 128

%define TABLE_1 ebx + 64
%define TABLE_2 ebx + 128
%define TABLE_3 ebx + 192
%define TABLE_4 ebx + 256
%define TABLE_5 ebx + 320
%define TABLE_6 ebx + 384
%define TABLE_7 ebx + 448

%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

%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)

;---added for sse2
%define tg_1_16_2 (TABLEF + 0)
%define tg_2_16_2 (TABLEF + 16)
%define tg_3_16_2 (TABLEF + 32)
%define cos_4_16_2 (TABLEF + 48)
%define ocos_4_16_2 (TABLEF + 64)
;---

%ifdef FORMAT_COFF
SECTION .data data
%else
SECTION .data data align=64
%endif


ALIGN 16

BITS_FRW_ACC equ 3							; 2 or 3 for accuracy
SHIFT_FRW_COL equ BITS_FRW_ACC
SHIFT_FRW_ROW equ (BITS_FRW_ACC + 17)
RND_FRW_ROW equ (1 << (SHIFT_FRW_ROW-1))

SHIFT_FRW_ROW_CLIP2	equ (4)
SHIFT_FRW_ROW_CLIP1	equ (SHIFT_FRW_ROW - SHIFT_FRW_ROW_CLIP2)

PASS1_BITS		equ	(2)
CONST_BITS		equ	(13)

one_corr		 dw 1, 1, 1, 1

;---added for sse2
align 16
one_corr_2		 dw 1, 1, 1, 1,
                         dw 1, 1, 1, 1
;---

r_frw_row		 dd RND_FRW_ROW, RND_FRW_ROW


tg_all_16		 dw 13036, 13036, 13036, 13036,		; tg * (2<<16) + 0.5
				 dw 27146, 27146, 27146, 27146,		; tg * (2<<16) + 0.5
				 dw -21746, -21746, -21746, -21746,	; tg * (2<<16) + 0.5
				 dw -19195, -19195, -19195, -19195,	; cos * (2<<16) + 0.5
				 dw 23170, 23170, 23170, 23170      ; cos * (2<<15) + 0.5

;---added for sse2
align 16
tg_all_16_2		 dw 13036, 13036, 13036, 13036,		; tg * (2<<16) + 0.5
                         dw 13036, 13036, 13036, 13036,        
				 dw 27146, 27146, 27146, 27146,		; tg * (2<<16) + 0.5
                                 dw 27146, 27146, 27146, 27146, 
				 dw -21746, -21746, -21746, -21746,	; tg * (2<<16) + 0.5
                                 dw -21746, -21746, -21746, -21746,
				 dw -19195, -19195, -19195, -19195,	; cos * (2<<16) + 0.5
                                 dw -19195, -19195, -19195, -19195,
				 dw 23170, 23170, 23170, 23170      ; cos * (2<<15) + 0.5
                                 dw 23170, 23170, 23170, 23170
;---

tab_frw_01234567 
				 ; row0
				 dw 16384, 16384, 21407, -8867,     ; w09 w01 w08 w00
				 dw 16384, 16384, 8867, -21407,     ; w13 w05 w12 w04
                 dw 16384, -16384, 8867, 21407,     ; w11 w03 w10 w02
                 dw -16384, 16384, -21407, -8867,   ; w15 w07 w14 w06
                 dw 22725, 12873, 19266, -22725,    ; w22 w20 w18 w16
                 dw 19266, 4520, -4520, -12873,     ; w23 w21 w19 w17
                 dw 12873, 4520, 4520, 19266,       ; w30 w28 w26 w24
                 dw -22725, 19266, -12873, -22725,  ; w31 w29 w27 w25

				 ; row1
                 dw 22725, 22725, 29692, -12299,    ; w09 w01 w08 w00
                 dw 22725, 22725, 12299, -29692,    ; w13 w05 w12 w04
                 dw 22725, -22725, 12299, 29692,    ; w11 w03 w10 w02
                 dw -22725, 22725, -29692, -12299,  ; w15 w07 w14 w06
                 dw 31521, 17855, 26722, -31521,    ; w22 w20 w18 w16
                 dw 26722, 6270, -6270, -17855,     ; w23 w21 w19 w17
                 dw 17855, 6270, 6270, 26722,       ; w30 w28 w26 w24
                 dw -31521, 26722, -17855, -31521,  ; w31 w29 w27 w25

				 ; row2
                 dw 21407, 21407, 27969, -11585,    ; w09 w01 w08 w00
                 dw 21407, 21407, 11585, -27969,    ; w13 w05 w12 w04
                 dw 21407, -21407, 11585, 27969,    ; w11 w03 w10 w02
                 dw -21407, 21407, -27969, -11585,	; w15 w07 w14 w06
                 dw 29692, 16819, 25172, -29692,    ; w22 w20 w18 w16
                 dw 25172, 5906, -5906, -16819,     ; w23 w21 w19 w17
                 dw 16819, 5906, 5906, 25172,       ; w30 w28 w26 w24
                 dw -29692, 25172, -16819, -29692,  ; w31 w29 w27 w25

				 ; row3
                 dw 19266, 19266, 25172, -10426,    ; w09 w01 w08 w00
                 dw 19266, 19266, 10426, -25172,    ; w13 w05 w12 w04
                 dw 19266, -19266, 10426, 25172,    ; w11 w03 w10 w02
                 dw -19266, 19266, -25172, -10426,  ; w15 w07 w14 w06 
                 dw 26722, 15137, 22654, -26722,    ; w22 w20 w18 w16
                 dw 22654, 5315, -5315, -15137,     ; w23 w21 w19 w17
                 dw 15137, 5315, 5315, 22654,       ; w30 w28 w26 w24
                 dw -26722, 22654, -15137, -26722,  ; w31 w29 w27 w25

				 ; row4
                 dw 16384, 16384, 21407, -8867,     ; w09 w01 w08 w00
                 dw 16384, 16384, 8867, -21407,     ; w13 w05 w12 w04
                 dw 16384, -16384, 8867, 21407,     ; w11 w03 w10 w02
                 dw -16384, 16384, -21407, -8867,   ; w15 w07 w14 w06
                 dw 22725, 12873, 19266, -22725,    ; w22 w20 w18 w16
                 dw 19266, 4520, -4520, -12873,     ; w23 w21 w19 w17
                 dw 12873, 4520, 4520, 19266,       ; w30 w28 w26 w24
                 dw -22725, 19266, -12873, -22725,  ; w31 w29 w27 w25 

				 ; row5
                 dw 19266, 19266, 25172, -10426,    ; w09 w01 w08 w00
                 dw 19266, 19266, 10426, -25172,    ; w13 w05 w12 w04
                 dw 19266, -19266, 10426, 25172,    ; w11 w03 w10 w02
                 dw -19266, 19266, -25172, -10426,  ; w15 w07 w14 w06
                 dw 26722, 15137, 22654, -26722,    ; w22 w20 w18 w16
                 dw 22654, 5315, -5315, -15137,     ; w23 w21 w19 w17
                 dw 15137, 5315, 5315, 22654,       ; w30 w28 w26 w24
                 dw -26722, 22654, -15137, -26722,  ; w31 w29 w27 w25

				 ; row6
                 dw 21407, 21407, 27969, -11585,    ; w09 w01 w08 w00
                 dw 21407, 21407, 11585, -27969,    ; w13 w05 w12 w04
                 dw 21407, -21407, 11585, 27969,    ; w11 w03 w10 w02
                 dw -21407, 21407, -27969, -11585,  ; w15 w07 w14 w06
                 dw 29692, 16819, 25172, -29692,    ; w22 w20 w18 w16
                 dw 25172, 5906, -5906, -16819,     ; w23 w21 w19 w17
                 dw 16819, 5906, 5906, 25172,       ; w30 w28 w26 w24
                 dw -29692, 25172, -16819, -29692,  ; w31 w29 w27 w25

				 ; row7
                 dw 22725, 22725, 29692, -12299,    ; w09 w01 w08 w00
                 dw 22725, 22725, 12299, -29692,    ; w13 w05 w12 w04
                 dw 22725, -22725, 12299, 29692,    ; w11 w03 w10 w02
                 dw -22725, 22725, -29692, -12299,  ; w15 w07 w14 w06
                 dw 31521, 17855, 26722, -31521,    ; w22 w20 w18 w16
                 dw 26722, 6270, -6270, -17855,     ; w23 w21 w19 w17
                 dw 17855, 6270, 6270, 26722,       ; w30 w28 w26 w24
                 dw -31521, 26722, -17855, -31521   ; w31 w29 w27 w25

align 128

FIX_1		dw 10703, 4433, 10703, 4433
		dw 10703, 4433, 10703, 4433

FIX_2		dw 4433, -10704, 4433, -10704
		dw 4433, -10704, 4433, -10704

rounder_5       dw 2+16,2+16,2+16,2+16,2+16,2+16,2+16,2+16
rounder_11      dd 1024,1024,1024,1024
rounder_18      dd 16384+131072,16384+131072,16384+131072,16384+131072

align 128

FIX_3		dw 6437, 2260, 6437, 2260
		dw 6437, 2260, 6437, 2260
		dw 11363, 9633, 11363, 9633
		dw 11363, 9633, 11363, 9633

FIX_4		dw -11362, -6436, -11362, -6436
		dw -11362, -6436, -11362, -6436
		dw 9633, -2259, 9633, -2259
		dw 9633, -2259, 9633, -2259

FIX_5		dw 2261, 9633, 2261, 9633
		dw 2261, 9633, 2261, 9633
		dw 6437, -11362, 6437, -11362
		dw 6437, -11362, 6437, -11362

FIX_6           dw 9633, -11363, 9633, -11363
                dw 9633, -11363, 9633, -11363
		dw 2260, -6436, 2260, -6436
		dw 2260, -6436, 2260, -6436

align 128
buffer		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0
		dw 0,0,0,0,0,0,0,0

SECTION .text

ALIGN 16


cglobal fdct_mmx
;;void fdct_mmx(short *block);
fdct_mmx:
	
	push ebx

    mov INP, dword [esp + 8]	; block

    mov TABLEF, tg_all_16
    mov OUT, INP

    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, qword [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