summaryrefslogtreecommitdiff
path: root/libavcodec/x86/h264_weight_10bit.asm
blob: 3b09e420d28da7ac5f3e5b6cb70de0ef00151558 (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
;*****************************************************************************
;* MMX/SSE2/AVX-optimized 10-bit H.264 weighted prediction code
;*****************************************************************************
;* Copyright (C) 2005-2011 x264 project
;*
;* Authors: Daniel Kang <daniel.d.kang@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_RODATA 32

pw_pixel_max: times 8 dw ((1 << 10)-1)
sq_1: dq 1
      dq 0

cextern pw_1

SECTION .text

;-----------------------------------------------------------------------------
; void h264_weight(uint8_t *dst, int stride, int height, int log2_denom,
;                  int weight, int offset);
;-----------------------------------------------------------------------------
%macro WEIGHT_PROLOGUE 0
.prologue:
    PROLOGUE 0,6,8
    movifnidn  r0, r0mp
    movifnidn r1d, r1m
    movifnidn r2d, r2m
    movifnidn r4d, r4m
    movifnidn r5d, r5m
%endmacro

%macro WEIGHT_SETUP 0
    mova       m0, [pw_1]
    movd       m2, r3m
    pslld      m0, m2       ; 1<<log2_denom
    SPLATW     m0, m0
    shl        r5, 19       ; *8, move to upper half of dword
    lea        r5, [r5+r4*2+0x10000]
    movd       m3, r5d      ; weight<<1 | 1+(offset<<(3))
    pshufd     m3, m3, 0
    mova       m4, [pw_pixel_max]
    paddw      m2, [sq_1]   ; log2_denom+1
%if notcpuflag(sse4)
    pxor       m7, m7
%endif
%endmacro

%macro WEIGHT_OP 1-2
%if %0==1
    mova        m5, [r0+%1]
    punpckhwd   m6, m5, m0
    punpcklwd   m5, m0
%else
    movq        m5, [r0+%1]
    movq        m6, [r0+%2]
    punpcklwd   m5, m0
    punpcklwd   m6, m0
%endif
    pmaddwd     m5, m3
    pmaddwd     m6, m3
    psrad       m5, m2
    psrad       m6, m2
%if cpuflag(sse4)
    packusdw    m5, m6
    pminsw      m5, m4
%else
    packssdw    m5, m6
    CLIPW       m5, m7, m4
%endif
%endmacro

%macro WEIGHT_FUNC_DBL 0
cglobal h264_weight_16_10
    WEIGHT_PROLOGUE
    WEIGHT_SETUP
.nextrow:
    WEIGHT_OP  0
    mova [r0   ], m5
    WEIGHT_OP 16
    mova [r0+16], m5
    add       r0, r1
    dec       r2d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
WEIGHT_FUNC_DBL
INIT_XMM sse4
WEIGHT_FUNC_DBL


%macro WEIGHT_FUNC_MM 0
cglobal h264_weight_8_10
    WEIGHT_PROLOGUE
    WEIGHT_SETUP
.nextrow:
    WEIGHT_OP   0
    mova     [r0], m5
    add        r0, r1
    dec        r2d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
WEIGHT_FUNC_MM
INIT_XMM sse4
WEIGHT_FUNC_MM


%macro WEIGHT_FUNC_HALF_MM 0
cglobal h264_weight_4_10
    WEIGHT_PROLOGUE
    sar         r2d, 1
    WEIGHT_SETUP
    lea         r3, [r1*2]
.nextrow:
    WEIGHT_OP    0, r1
    movh      [r0], m5
    movhps [r0+r1], m5
    add         r0, r3
    dec         r2d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
WEIGHT_FUNC_HALF_MM
INIT_XMM sse4
WEIGHT_FUNC_HALF_MM


;-----------------------------------------------------------------------------
; void h264_biweight(uint8_t *dst, uint8_t *src, int stride, int height,
;                    int log2_denom, int weightd, int weights, int offset);
;-----------------------------------------------------------------------------
%if ARCH_X86_32
DECLARE_REG_TMP 3
%else
DECLARE_REG_TMP 7
%endif

%macro BIWEIGHT_PROLOGUE 0
.prologue:
    PROLOGUE 0,8,8
    movifnidn  r0, r0mp
    movifnidn  r1, r1mp
    movifnidn r2d, r2m
    movifnidn r5d, r5m
    movifnidn r6d, r6m
    movifnidn t0d, r7m
%endmacro

%macro BIWEIGHT_SETUP 0
    lea        t0, [t0*4+1] ; (offset<<2)+1
    or         t0, 1
    shl        r6, 16
    or         r5, r6
    movd       m4, r5d      ; weightd | weights
    movd       m5, t0d      ; (offset+1)|1
    movd       m6, r4m      ; log2_denom
    pslld      m5, m6       ; (((offset<<2)+1)|1)<<log2_denom
    paddd      m6, [sq_1]
    pshufd     m4, m4, 0
    pshufd     m5, m5, 0
    mova       m3, [pw_pixel_max]
    movifnidn r3d, r3m
%if notcpuflag(sse4)
    pxor       m7, m7
%endif
%endmacro

%macro BIWEIGHT 1-2
%if %0==1
    mova       m0, [r0+%1]
    mova       m1, [r1+%1]
    punpckhwd  m2, m0, m1
    punpcklwd  m0, m1
%else
    movq       m0, [r0+%1]
    movq       m1, [r1+%1]
    punpcklwd  m0, m1
    movq       m2, [r0+%2]
    movq       m1, [r1+%2]
    punpcklwd  m2, m1
%endif
    pmaddwd    m0, m4
    pmaddwd    m2, m4
    paddd      m0, m5
    paddd      m2, m5
    psrad      m0, m6
    psrad      m2, m6
%if cpuflag(sse4)
    packusdw   m0, m2
    pminsw     m0, m3
%else
    packssdw   m0, m2
    CLIPW      m0, m7, m3
%endif
%endmacro

%macro BIWEIGHT_FUNC_DBL 0
cglobal h264_biweight_16_10
    BIWEIGHT_PROLOGUE
    BIWEIGHT_SETUP
.nextrow:
    BIWEIGHT   0
    mova [r0   ], m0
    BIWEIGHT  16
    mova [r0+16], m0
    add       r0, r2
    add       r1, r2
    dec       r3d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
BIWEIGHT_FUNC_DBL
INIT_XMM sse4
BIWEIGHT_FUNC_DBL

%macro BIWEIGHT_FUNC 0
cglobal h264_biweight_8_10
    BIWEIGHT_PROLOGUE
    BIWEIGHT_SETUP
.nextrow:
    BIWEIGHT  0
    mova   [r0], m0
    add      r0, r2
    add      r1, r2
    dec      r3d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
BIWEIGHT_FUNC
INIT_XMM sse4
BIWEIGHT_FUNC

%macro BIWEIGHT_FUNC_HALF 0
cglobal h264_biweight_4_10
    BIWEIGHT_PROLOGUE
    BIWEIGHT_SETUP
    sar        r3d, 1
    lea        r4, [r2*2]
.nextrow:
    BIWEIGHT     0, r2
    movh   [r0   ], m0
    movhps [r0+r2], m0
    add         r0, r4
    add         r1, r4
    dec         r3d
    jnz .nextrow
    REP_RET
%endmacro

INIT_XMM sse2
BIWEIGHT_FUNC_HALF
INIT_XMM sse4
BIWEIGHT_FUNC_HALF