aboutsummaryrefslogtreecommitdiff
path: root/libavcodec/x86/h264_weight.asm
blob: 646acdffe30b0f8c6f39d3b0a2fef8f178ad2b62 (plain)
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
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
;*****************************************************************************
;* SSE2-optimized weighted prediction code
;*****************************************************************************
;* Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt
;* Copyright (C) 2010 Eli Friedman <eli.friedman@gmail.com>
;*
;* This file is part of Libav.
;*
;* Libav 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.
;*
;* Libav 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 Libav; 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 .text

;-----------------------------------------------------------------------------
; biweight pred:
;
; void h264_biweight_16_sse2(uint8_t *dst, uint8_t *src, int stride,
;                            int height, int log2_denom, int weightd,
;                            int weights, int offset);
; and
; void h264_weight_16_sse2(uint8_t *dst, int stride, int height,
;                          int log2_denom, int weight, int offset);
;-----------------------------------------------------------------------------

%macro WEIGHT_SETUP 0
    add        r5, r5
    inc        r5
    movd       m3, r4d
    movd       m5, r5d
    movd       m6, r3d
    pslld      m5, m6
    psrld      m5, 1
%if mmsize == 16
    pshuflw    m3, m3, 0
    pshuflw    m5, m5, 0
    punpcklqdq m3, m3
    punpcklqdq m5, m5
%else
    pshufw     m3, m3, 0
    pshufw     m5, m5, 0
%endif
    pxor       m7, m7
%endmacro

%macro WEIGHT_OP 2
    movh          m0, [r0+%1]
    movh          m1, [r0+%2]
    punpcklbw     m0, m7
    punpcklbw     m1, m7
    pmullw        m0, m3
    pmullw        m1, m3
    paddsw        m0, m5
    paddsw        m1, m5
    psraw         m0, m6
    psraw         m1, m6
    packuswb      m0, m1
%endmacro

INIT_MMX mmxext
cglobal h264_weight_16, 6, 6, 0
    WEIGHT_SETUP
.nextrow:
    WEIGHT_OP 0,  4
    mova     [r0  ], m0
    WEIGHT_OP 8, 12
    mova     [r0+8], m0
    add        r0, r1
    dec        r2d
    jnz .nextrow
    REP_RET

%macro WEIGHT_FUNC_MM 2
cglobal h264_weight_%1, 6, 6, %2
    WEIGHT_SETUP
.nextrow:
    WEIGHT_OP 0, mmsize/2
    mova     [r0], m0
    add        r0, r1
    dec        r2d
    jnz .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
WEIGHT_FUNC_MM  8, 0
INIT_XMM sse2
WEIGHT_FUNC_MM 16, 8

%macro WEIGHT_FUNC_HALF_MM 2
cglobal h264_weight_%1, 6, 6, %2
    WEIGHT_SETUP
    sar       r2d, 1
    lea        r3, [r1*2]
.nextrow:
    WEIGHT_OP 0, r1
    movh     [r0], m0
%if mmsize == 16
    movhps   [r0+r1], m0
%else
    psrlq      m0, 32
    movh     [r0+r1], m0
%endif
    add        r0, r3
    dec        r2d
    jnz .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
WEIGHT_FUNC_HALF_MM 4, 0
INIT_XMM sse2
WEIGHT_FUNC_HALF_MM 8, 8

%macro BIWEIGHT_SETUP 0
%if ARCH_X86_64
%define off_regd r7d
%else
%define off_regd r3d
%endif
    mov  off_regd, r7m
    add  off_regd, 1
    or   off_regd, 1
    add        r4, 1
%if cpuflag(ssse3)
    movd       m4, r5d
    movd       m0, r6d
%else
    movd       m3, r5d
    movd       m4, r6d
%endif
    movd       m5, off_regd
    movd       m6, r4d
    pslld      m5, m6
    psrld      m5, 1
%if cpuflag(ssse3)
    punpcklbw  m4, m0
    pshuflw    m4, m4, 0
    pshuflw    m5, m5, 0
    punpcklqdq m4, m4
    punpcklqdq m5, m5

%else
%if mmsize == 16
    pshuflw    m3, m3, 0
    pshuflw    m4, m4, 0
    pshuflw    m5, m5, 0
    punpcklqdq m3, m3
    punpcklqdq m4, m4
    punpcklqdq m5, m5
%else
    pshufw     m3, m3, 0
    pshufw     m4, m4, 0
    pshufw     m5, m5, 0
%endif
    pxor       m7, m7
%endif
%endmacro

%macro BIWEIGHT_STEPA 3
    movh       m%1, [r0+%3]
    movh       m%2, [r1+%3]
    punpcklbw  m%1, m7
    punpcklbw  m%2, m7
    pmullw     m%1, m3
    pmullw     m%2, m4
    paddsw     m%1, m%2
%endmacro

%macro BIWEIGHT_STEPB 0
    paddsw     m0, m5
    paddsw     m1, m5
    psraw      m0, m6
    psraw      m1, m6
    packuswb   m0, m1
%endmacro

INIT_MMX mmxext
cglobal h264_biweight_16, 7, 8, 0
    BIWEIGHT_SETUP
    movifnidn r3d, r3m
.nextrow:
    BIWEIGHT_STEPA 0, 1, 0
    BIWEIGHT_STEPA 1, 2, 4
    BIWEIGHT_STEPB
    mova       [r0], m0
    BIWEIGHT_STEPA 0, 1, 8
    BIWEIGHT_STEPA 1, 2, 12
    BIWEIGHT_STEPB
    mova     [r0+8], m0
    add        r0, r2
    add        r1, r2
    dec        r3d
    jnz .nextrow
    REP_RET

%macro BIWEIGHT_FUNC_MM 2
cglobal h264_biweight_%1, 7, 8, %2
    BIWEIGHT_SETUP
    movifnidn r3d, r3m
.nextrow:
    BIWEIGHT_STEPA 0, 1, 0
    BIWEIGHT_STEPA 1, 2, mmsize/2
    BIWEIGHT_STEPB
    mova       [r0], m0
    add        r0, r2
    add        r1, r2
    dec        r3d
    jnz .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
BIWEIGHT_FUNC_MM  8, 0
INIT_XMM sse2
BIWEIGHT_FUNC_MM 16, 8

%macro BIWEIGHT_FUNC_HALF_MM 2
cglobal h264_biweight_%1, 7, 8, %2
    BIWEIGHT_SETUP
    movifnidn r3d, r3m
    sar        r3, 1
    lea        r4, [r2*2]
.nextrow:
    BIWEIGHT_STEPA 0, 1, 0
    BIWEIGHT_STEPA 1, 2, r2
    BIWEIGHT_STEPB
    movh       [r0], m0
%if mmsize == 16
    movhps     [r0+r2], m0
%else
    psrlq      m0, 32
    movh       [r0+r2], m0
%endif
    add        r0, r4
    add        r1, r4
    dec        r3d
    jnz .nextrow
    REP_RET
%endmacro

INIT_MMX mmxext
BIWEIGHT_FUNC_HALF_MM 4, 0
INIT_XMM sse2
BIWEIGHT_FUNC_HALF_MM 8, 8

%macro BIWEIGHT_SSSE3_OP 0
    pmaddubsw  m0, m4
    pmaddubsw  m2, m4
    paddsw     m0, m5
    paddsw     m2, m5
    psraw      m0, m6
    psraw      m2, m6
    packuswb   m0, m2
%endmacro

INIT_XMM ssse3
cglobal h264_biweight_16, 7, 8, 8
    BIWEIGHT_SETUP
    movifnidn r3d, r3m

.nextrow:
    movh       m0, [r0]
    movh       m2, [r0+8]
    movh       m3, [r1+8]
    punpcklbw  m0, [r1]
    punpcklbw  m2, m3
    BIWEIGHT_SSSE3_OP
    mova       [r0], m0
    add        r0, r2
    add        r1, r2
    dec        r3d
    jnz .nextrow
    REP_RET

INIT_XMM ssse3
cglobal h264_biweight_8, 7, 8, 8
    BIWEIGHT_SETUP
    movifnidn r3d, r3m
    sar        r3, 1
    lea        r4, [r2*2]

.nextrow:
    movh       m0, [r0]
    movh       m1, [r1]
    movh       m2, [r0+r2]
    movh       m3, [r1+r2]
    punpcklbw  m0, m1
    punpcklbw  m2, m3
    BIWEIGHT_SSSE3_OP
    movh       [r0], m0
    movhps     [r0+r2], m0
    add        r0, r4
    add        r1, r4
    dec        r3d
    jnz .nextrow
    REP_RET