aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/x86/rv34dsp.asm
blob: 5568ddfdf8217828b83a57f7e1cbcdfbf18f655f (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
;******************************************************************************
;* MMX/SSE2-optimized functions for the RV30 and RV40 decoders
;* Copyright (C) 2012 Christophe Gisquet <christophe.gisquet@gmail.com>
;*
;* This file is part of FFmpeg.
;*
;* FFmpeg is free software; you can redistribute it and/or
;* modify it under the terms of the GNU Lesser General Public
;* License as published by the Free Software Foundation; either
;* version 2.1 of the License, or (at your option) any later version.
;*
;* FFmpeg 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
;* Lesser General Public License for more details.
;*
;* You should have received a copy of the GNU Lesser General Public
;* License along with FFmpeg; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************

%include "libavutil/x86/x86util.asm"

SECTION_RODATA
pw_row_coeffs:  times 4 dw 13
                times 4 dw 17
                times 4 dw  7
pd_512: times 2 dd 0x200
pw_col_coeffs:  dw 13,  13,  13, -13
                dw 17,   7,   7, -17
                dw 13, -13,  13,  13
                dw -7,  17, -17,  -7

SECTION .text

%macro IDCT_DC_NOROUND 1
    imul   %1, 13*13*3
    sar    %1, 11
%endmacro

%macro IDCT_DC_ROUND 1
    imul   %1, 13*13
    add    %1, 0x200
    sar    %1, 10
%endmacro

INIT_MMX mmxext
cglobal rv34_idct_dc_noround, 1, 2, 0
    movsx   r1, word [r0]
    IDCT_DC_NOROUND r1
    movd    m0, r1d
    pshufw  m0, m0, 0
    movq    [r0+ 0], m0
    movq    [r0+ 8], m0
    movq    [r0+16], m0
    movq    [r0+24], m0
    REP_RET

; ff_rv34_idct_dc_add_mmx(uint8_t *dst, int stride, int dc);
%if ARCH_X86_32
INIT_MMX mmx
cglobal rv34_idct_dc_add, 3, 3
    ; calculate DC
    IDCT_DC_ROUND r2
    pxor       m1, m1
    movd       m0, r2d
    psubw      m1, m0
    packuswb   m0, m0
    packuswb   m1, m1
    punpcklbw  m0, m0
    punpcklbw  m1, m1
    punpcklwd  m0, m0
    punpcklwd  m1, m1

    ; add DC
    lea        r2, [r0+r1*2]
    movh       m2, [r0]
    movh       m3, [r0+r1]
    movh       m4, [r2]
    movh       m5, [r2+r1]
    paddusb    m2, m0
    paddusb    m3, m0
    paddusb    m4, m0
    paddusb    m5, m0
    psubusb    m2, m1
    psubusb    m3, m1
    psubusb    m4, m1
    psubusb    m5, m1
    movh       [r0], m2
    movh       [r0+r1], m3
    movh       [r2], m4
    movh       [r2+r1], m5
    RET
%endif

; Load coeffs and perform row transform
; Output: coeffs in mm[0467], rounder in mm5
%macro ROW_TRANSFORM  1
    pxor        mm7, mm7
    mova        mm0, [%1+ 0*8]
    mova        mm1, [%1+ 1*8]
    mova        mm2, [%1+ 2*8]
    mova        mm3, [%1+ 3*8]
    mova  [%1+ 0*8], mm7
    mova  [%1+ 1*8], mm7
    mova  [%1+ 2*8], mm7
    mova  [%1+ 3*8], mm7
    mova        mm4, mm0
    mova        mm6, [pw_row_coeffs+ 0]
    paddsw      mm0, mm2                ; b0 + b2
    psubsw      mm4, mm2                ; b0 - b2
    pmullw      mm0, mm6                ; *13 = z0
    pmullw      mm4, mm6                ; *13 = z1
    mova        mm5, mm1
    pmullw      mm1, [pw_row_coeffs+ 8] ; b1*17
    pmullw      mm5, [pw_row_coeffs+16] ; b1* 7
    mova        mm7, mm3
    pmullw      mm3, [pw_row_coeffs+ 8] ; b3*17
    pmullw      mm7, [pw_row_coeffs+16] ; b3* 7
    paddsw      mm1, mm7                ; z3 = b1*17 + b3* 7
    psubsw      mm5, mm3                ; z2 = b1* 7 - b3*17
    mova        mm7, mm0
    mova        mm6, mm4
    paddsw      mm0, mm1                ; z0 + z3
    psubsw      mm7, mm1                ; z0 - z3
    paddsw      mm4, mm5                ; z1 + z2
    psubsw      mm6, mm5                ; z1 - z2
    mova        mm5, [pd_512]           ; 0x200
%endmacro

; ff_rv34_idct_add_mmxext(uint8_t *dst, ptrdiff_t stride, int16_t *block);
%macro COL_TRANSFORM  4
    pshufw      mm3, %2, 0xDD        ; col. 1,3,1,3
    pshufw       %2, %2, 0x88        ; col. 0,2,0,2
    pmaddwd      %2, %3              ; 13*c0+13*c2 | 13*c0-13*c2 = z0 | z1
    pmaddwd     mm3, %4              ; 17*c1+ 7*c3 |  7*c1-17*c3 = z3 | z2
    paddd        %2, mm5
    pshufw      mm1,  %2, 01001110b  ;    z1 | z0
    pshufw      mm2, mm3, 01001110b  ;    z2 | z3
    paddd        %2, mm3             ; z0+z3 | z1+z2
    psubd       mm1, mm2             ; z1-z2 | z0-z3
    movd        mm3, %1
    psrad        %2, 10
    pxor        mm2, mm2
    psrad       mm1, 10
    punpcklbw   mm3, mm2
    packssdw     %2, mm1
    paddw        %2, mm3
    packuswb     %2, %2
    movd         %1, %2
%endmacro
INIT_MMX mmxext
cglobal rv34_idct_add, 3,3,0, d, s, b
    ROW_TRANSFORM       bq
    COL_TRANSFORM     [dq], mm0, [pw_col_coeffs+ 0], [pw_col_coeffs+ 8]
    mova               mm0, [pw_col_coeffs+ 0]
    COL_TRANSFORM  [dq+sq], mm4, mm0, [pw_col_coeffs+ 8]
    mova               mm4, [pw_col_coeffs+ 8]
    lea                 dq, [dq + 2*sq]
    COL_TRANSFORM     [dq], mm6, mm0, mm4
    COL_TRANSFORM  [dq+sq], mm7, mm0, mm4
    ret

; ff_rv34_idct_dc_add_sse4(uint8_t *dst, int stride, int dc);
%macro RV34_IDCT_DC_ADD 0
cglobal rv34_idct_dc_add, 3, 3, 6
    ; load data
    IDCT_DC_ROUND r2
    pxor       m1, m1

    ; calculate DC
    movd       m0, r2d
    lea        r2, [r0+r1*2]
    movd       m2, [r0]
    movd       m3, [r0+r1]
    pshuflw    m0, m0, 0
    movd       m4, [r2]
    movd       m5, [r2+r1]
    punpcklqdq m0, m0
    punpckldq  m2, m3
    punpckldq  m4, m5
    punpcklbw  m2, m1
    punpcklbw  m4, m1
    paddw      m2, m0
    paddw      m4, m0
    packuswb   m2, m4
    movd      [r0], m2
%if cpuflag(sse4)
    pextrd [r0+r1], m2, 1
    pextrd    [r2], m2, 2
    pextrd [r2+r1], m2, 3
%else
    psrldq     m2, 4
    movd   [r0+r1], m2
    psrldq     m2, 4
    movd      [r2], m2
    psrldq     m2, 4
    movd   [r2+r1], m2
%endif
    RET
%endmacro

INIT_XMM sse2
RV34_IDCT_DC_ADD
INIT_XMM sse4
RV34_IDCT_DC_ADD