summaryrefslogtreecommitdiffstats
path: root/media/libvpx/vpx_dsp/x86/sad_sse2.asm
blob: c6a829dc21e3630dcce482cb41ef3053644905e1 (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
;
;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
;  Use of this source code is governed by a BSD-style license
;  that can be found in the LICENSE file in the root of the source
;  tree. An additional intellectual property rights grant can be found
;  in the file PATENTS.  All contributing project authors may
;  be found in the AUTHORS file in the root of the source tree.
;

%define program_name vpx

%include "third_party/x86inc/x86inc.asm"

SECTION .text

%macro SAD_FN 4
%if %4 == 0
%if %3 == 5
cglobal sad%1x%2, 4, %3, 5, src, src_stride, ref, ref_stride, n_rows
%else ; %3 == 7
cglobal sad%1x%2, 4, %3, 5, src, src_stride, ref, ref_stride, \
                            src_stride3, ref_stride3, n_rows
%endif ; %3 == 5/7
%else ; avg
%if %3 == 5
cglobal sad%1x%2_avg, 5, 1 + %3, 5, src, src_stride, ref, ref_stride, \
                                    second_pred, n_rows
%else ; %3 == 7
cglobal sad%1x%2_avg, 5, ARCH_X86_64 + %3, 5, src, src_stride, \
                                              ref, ref_stride, \
                                              second_pred, \
                                              src_stride3, ref_stride3
%if ARCH_X86_64
%define n_rowsd r7d
%else ; x86-32
%define n_rowsd dword r0m
%endif ; x86-32/64
%endif ; %3 == 5/7
%endif ; avg/sad
  movsxdifnidn src_strideq, src_strided
  movsxdifnidn ref_strideq, ref_strided
%if %3 == 7
  lea         src_stride3q, [src_strideq*3]
  lea         ref_stride3q, [ref_strideq*3]
%endif ; %3 == 7
%endmacro

; unsigned int vpx_sad64x64_sse2(uint8_t *src, int src_stride,
;                                uint8_t *ref, int ref_stride);
%macro SAD64XN 1-2 0
  SAD_FN 64, %1, 5, %2
  mov              n_rowsd, %1
  pxor                  m0, m0
.loop:
  movu                  m1, [refq]
  movu                  m2, [refq+16]
  movu                  m3, [refq+32]
  movu                  m4, [refq+48]
%if %2 == 1
  pavgb                 m1, [second_predq+mmsize*0]
  pavgb                 m2, [second_predq+mmsize*1]
  pavgb                 m3, [second_predq+mmsize*2]
  pavgb                 m4, [second_predq+mmsize*3]
  lea         second_predq, [second_predq+mmsize*4]
%endif
  psadbw                m1, [srcq]
  psadbw                m2, [srcq+16]
  psadbw                m3, [srcq+32]
  psadbw                m4, [srcq+48]
  paddd                 m1, m2
  paddd                 m3, m4
  add                 refq, ref_strideq
  paddd                 m0, m1
  add                 srcq, src_strideq
  paddd                 m0, m3
  dec              n_rowsd
  jg .loop

  movhlps               m1, m0
  paddd                 m0, m1
  movd                 eax, m0
  RET
%endmacro

INIT_XMM sse2
SAD64XN 64 ; sad64x64_sse2
SAD64XN 32 ; sad64x32_sse2
SAD64XN 64, 1 ; sad64x64_avg_sse2
SAD64XN 32, 1 ; sad64x32_avg_sse2

; unsigned int vpx_sad32x32_sse2(uint8_t *src, int src_stride,
;                                uint8_t *ref, int ref_stride);
%macro SAD32XN 1-2 0
  SAD_FN 32, %1, 5, %2
  mov              n_rowsd, %1/2
  pxor                  m0, m0
.loop:
  movu                  m1, [refq]
  movu                  m2, [refq+16]
  movu                  m3, [refq+ref_strideq]
  movu                  m4, [refq+ref_strideq+16]
%if %2 == 1
  pavgb                 m1, [second_predq+mmsize*0]
  pavgb                 m2, [second_predq+mmsize*1]
  pavgb                 m3, [second_predq+mmsize*2]
  pavgb                 m4, [second_predq+mmsize*3]
  lea         second_predq, [second_predq+mmsize*4]
%endif
  psadbw                m1, [srcq]
  psadbw                m2, [srcq+16]
  psadbw                m3, [srcq+src_strideq]
  psadbw                m4, [srcq+src_strideq+16]
  paddd                 m1, m2
  paddd                 m3, m4
  lea                 refq, [refq+ref_strideq*2]
  paddd                 m0, m1
  lea                 srcq, [srcq+src_strideq*2]
  paddd                 m0, m3
  dec              n_rowsd
  jg .loop

  movhlps               m1, m0
  paddd                 m0, m1
  movd                 eax, m0
  RET
%endmacro

INIT_XMM sse2
SAD32XN 64 ; sad32x64_sse2
SAD32XN 32 ; sad32x32_sse2
SAD32XN 16 ; sad32x16_sse2
SAD32XN 64, 1 ; sad32x64_avg_sse2
SAD32XN 32, 1 ; sad32x32_avg_sse2
SAD32XN 16, 1 ; sad32x16_avg_sse2

; unsigned int vpx_sad16x{8,16}_sse2(uint8_t *src, int src_stride,
;                                    uint8_t *ref, int ref_stride);
%macro SAD16XN 1-2 0
  SAD_FN 16, %1, 7, %2
  mov              n_rowsd, %1/4
  pxor                  m0, m0

.loop:
  movu                  m1, [refq]
  movu                  m2, [refq+ref_strideq]
  movu                  m3, [refq+ref_strideq*2]
  movu                  m4, [refq+ref_stride3q]
%if %2 == 1
  pavgb                 m1, [second_predq+mmsize*0]
  pavgb                 m2, [second_predq+mmsize*1]
  pavgb                 m3, [second_predq+mmsize*2]
  pavgb                 m4, [second_predq+mmsize*3]
  lea         second_predq, [second_predq+mmsize*4]
%endif
  psadbw                m1, [srcq]
  psadbw                m2, [srcq+src_strideq]
  psadbw                m3, [srcq+src_strideq*2]
  psadbw                m4, [srcq+src_stride3q]
  paddd                 m1, m2
  paddd                 m3, m4
  lea                 refq, [refq+ref_strideq*4]
  paddd                 m0, m1
  lea                 srcq, [srcq+src_strideq*4]
  paddd                 m0, m3
  dec              n_rowsd
  jg .loop

  movhlps               m1, m0
  paddd                 m0, m1
  movd                 eax, m0
  RET
%endmacro

INIT_XMM sse2
SAD16XN 32 ; sad16x32_sse2
SAD16XN 16 ; sad16x16_sse2
SAD16XN  8 ; sad16x8_sse2
SAD16XN 32, 1 ; sad16x32_avg_sse2
SAD16XN 16, 1 ; sad16x16_avg_sse2
SAD16XN  8, 1 ; sad16x8_avg_sse2

; unsigned int vpx_sad8x{8,16}_sse2(uint8_t *src, int src_stride,
;                                   uint8_t *ref, int ref_stride);
%macro SAD8XN 1-2 0
  SAD_FN 8, %1, 7, %2
  mov              n_rowsd, %1/4
  pxor                  m0, m0

.loop:
  movh                  m1, [refq]
  movhps                m1, [refq+ref_strideq]
  movh                  m2, [refq+ref_strideq*2]
  movhps                m2, [refq+ref_stride3q]
%if %2 == 1
  pavgb                 m1, [second_predq+mmsize*0]
  pavgb                 m2, [second_predq+mmsize*1]
  lea         second_predq, [second_predq+mmsize*2]
%endif
  movh                  m3, [srcq]
  movhps                m3, [srcq+src_strideq]
  movh                  m4, [srcq+src_strideq*2]
  movhps                m4, [srcq+src_stride3q]
  psadbw                m1, m3
  psadbw                m2, m4
  lea                 refq, [refq+ref_strideq*4]
  paddd                 m0, m1
  lea                 srcq, [srcq+src_strideq*4]
  paddd                 m0, m2
  dec              n_rowsd
  jg .loop

  movhlps               m1, m0
  paddd                 m0, m1
  movd                 eax, m0
  RET
%endmacro

INIT_XMM sse2
SAD8XN 16 ; sad8x16_sse2
SAD8XN  8 ; sad8x8_sse2
SAD8XN  4 ; sad8x4_sse2
SAD8XN 16, 1 ; sad8x16_avg_sse2
SAD8XN  8, 1 ; sad8x8_avg_sse2
SAD8XN  4, 1 ; sad8x4_avg_sse2

; unsigned int vpx_sad4x{4, 8}_sse(uint8_t *src, int src_stride,
;                                  uint8_t *ref, int ref_stride);
%macro SAD4XN 1-2 0
  SAD_FN 4, %1, 7, %2
  mov              n_rowsd, %1/4
  pxor                  m0, m0

.loop:
  movd                  m1, [refq]
  movd                  m2, [refq+ref_strideq]
  movd                  m3, [refq+ref_strideq*2]
  movd                  m4, [refq+ref_stride3q]
  punpckldq             m1, m2
  punpckldq             m3, m4
%if %2 == 1
  pavgb                 m1, [second_predq+mmsize*0]
  pavgb                 m3, [second_predq+mmsize*1]
  lea         second_predq, [second_predq+mmsize*2]
%endif
  movd                  m2, [srcq]
  movd                  m5, [srcq+src_strideq]
  movd                  m4, [srcq+src_strideq*2]
  movd                  m6, [srcq+src_stride3q]
  punpckldq             m2, m5
  punpckldq             m4, m6
  psadbw                m1, m2
  psadbw                m3, m4
  lea                 refq, [refq+ref_strideq*4]
  paddd                 m0, m1
  lea                 srcq, [srcq+src_strideq*4]
  paddd                 m0, m3
  dec              n_rowsd
  jg .loop

  movd                 eax, m0
  RET
%endmacro

INIT_MMX sse
SAD4XN  8 ; sad4x8_sse
SAD4XN  4 ; sad4x4_sse
SAD4XN  8, 1 ; sad4x8_avg_sse
SAD4XN  4, 1 ; sad4x4_avg_sse