summaryrefslogtreecommitdiffstats
path: root/media/libaom/src/aom_dsp/x86/sad4d_sse2.asm
blob: 55a856985ad6180900900e240109cfefdf05d11c (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
;
; Copyright (c) 2016, Alliance for Open Media. All rights reserved
;
; This source code is subject to the terms of the BSD 2 Clause License and
; the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
; was not distributed with this source code in the LICENSE file, you can
; obtain it at www.aomedia.org/license/software. If the Alliance for Open
; Media Patent License 1.0 was not distributed with this source code in the
; PATENTS file, you can obtain it at www.aomedia.org/license/patent.
;

;

%include "third_party/x86inc/x86inc.asm"

SECTION .text

; PROCESS_4x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_4x2x4 5-6 0
  movd                  m0, [srcq +%2]
%if %1 == 1
  movd                  m6, [ref1q+%3]
  movd                  m4, [ref2q+%3]
  movd                  m7, [ref3q+%3]
  movd                  m5, [ref4q+%3]
  movd                  m1, [srcq +%4]
  movd                  m2, [ref1q+%5]
  punpckldq             m0, m1
  punpckldq             m6, m2
  movd                  m1, [ref2q+%5]
  movd                  m2, [ref3q+%5]
  movd                  m3, [ref4q+%5]
  punpckldq             m4, m1
  punpckldq             m7, m2
  punpckldq             m5, m3
  movlhps               m0, m0
  movlhps               m6, m4
  movlhps               m7, m5
  psadbw                m6, m0
  psadbw                m7, m0
%else
  movd                  m1, [ref1q+%3]
  movd                  m5, [ref1q+%5]
  movd                  m2, [ref2q+%3]
  movd                  m4, [ref2q+%5]
  punpckldq             m1, m5
  punpckldq             m2, m4
  movd                  m3, [ref3q+%3]
  movd                  m5, [ref3q+%5]
  punpckldq             m3, m5
  movd                  m4, [ref4q+%3]
  movd                  m5, [ref4q+%5]
  punpckldq             m4, m5
  movd                  m5, [srcq +%4]
  punpckldq             m0, m5
  movlhps               m0, m0
  movlhps               m1, m2
  movlhps               m3, m4
  psadbw                m1, m0
  psadbw                m3, m0
  paddd                 m6, m1
  paddd                 m7, m3
%endif
%if %6 == 1
  lea                 srcq, [srcq +src_strideq*2]
  lea                ref1q, [ref1q+ref_strideq*2]
  lea                ref2q, [ref2q+ref_strideq*2]
  lea                ref3q, [ref3q+ref_strideq*2]
  lea                ref4q, [ref4q+ref_strideq*2]
%endif
%endmacro

; PROCESS_8x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_8x2x4 5-6 0
  movh                  m0, [srcq +%2]
%if %1 == 1
  movh                  m4, [ref1q+%3]
  movh                  m5, [ref2q+%3]
  movh                  m6, [ref3q+%3]
  movh                  m7, [ref4q+%3]
  movhps                m0, [srcq +%4]
  movhps                m4, [ref1q+%5]
  movhps                m5, [ref2q+%5]
  movhps                m6, [ref3q+%5]
  movhps                m7, [ref4q+%5]
  psadbw                m4, m0
  psadbw                m5, m0
  psadbw                m6, m0
  psadbw                m7, m0
%else
  movh                  m1, [ref1q+%3]
  movh                  m2, [ref2q+%3]
  movh                  m3, [ref3q+%3]
  movhps                m0, [srcq +%4]
  movhps                m1, [ref1q+%5]
  movhps                m2, [ref2q+%5]
  movhps                m3, [ref3q+%5]
  psadbw                m1, m0
  psadbw                m2, m0
  psadbw                m3, m0
  paddd                 m4, m1
  movh                  m1, [ref4q+%3]
  movhps                m1, [ref4q+%5]
  paddd                 m5, m2
  paddd                 m6, m3
  psadbw                m1, m0
  paddd                 m7, m1
%endif
%if %6 == 1
  lea                 srcq, [srcq +src_strideq*2]
  lea                ref1q, [ref1q+ref_strideq*2]
  lea                ref2q, [ref2q+ref_strideq*2]
  lea                ref3q, [ref3q+ref_strideq*2]
  lea                ref4q, [ref4q+ref_strideq*2]
%endif
%endmacro

; PROCESS_16x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_16x2x4 5-6 0
  ; 1st 16 px
  mova                  m0, [srcq +%2]
%if %1 == 1
  movu                  m4, [ref1q+%3]
  movu                  m5, [ref2q+%3]
  movu                  m6, [ref3q+%3]
  movu                  m7, [ref4q+%3]
  psadbw                m4, m0
  psadbw                m5, m0
  psadbw                m6, m0
  psadbw                m7, m0
%else
  movu                  m1, [ref1q+%3]
  movu                  m2, [ref2q+%3]
  movu                  m3, [ref3q+%3]
  psadbw                m1, m0
  psadbw                m2, m0
  psadbw                m3, m0
  paddd                 m4, m1
  movu                  m1, [ref4q+%3]
  paddd                 m5, m2
  paddd                 m6, m3
  psadbw                m1, m0
  paddd                 m7, m1
%endif

  ; 2nd 16 px
  mova                  m0, [srcq +%4]
  movu                  m1, [ref1q+%5]
  movu                  m2, [ref2q+%5]
  movu                  m3, [ref3q+%5]
  psadbw                m1, m0
  psadbw                m2, m0
  psadbw                m3, m0
  paddd                 m4, m1
  movu                  m1, [ref4q+%5]
  paddd                 m5, m2
  paddd                 m6, m3
%if %6 == 1
  lea                 srcq, [srcq +src_strideq*2]
  lea                ref1q, [ref1q+ref_strideq*2]
  lea                ref2q, [ref2q+ref_strideq*2]
  lea                ref3q, [ref3q+ref_strideq*2]
  lea                ref4q, [ref4q+ref_strideq*2]
%endif
  psadbw                m1, m0
  paddd                 m7, m1
%endmacro

; PROCESS_32x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_32x2x4 5-6 0
  PROCESS_16x2x4 %1, %2, %3, %2 + 16, %3 + 16
  PROCESS_16x2x4  0, %4, %5, %4 + 16, %5 + 16, %6
%endmacro

; PROCESS_64x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_64x2x4 5-6 0
  PROCESS_32x2x4 %1, %2, %3, %2 + 32, %3 + 32
  PROCESS_32x2x4  0, %4, %5, %4 + 32, %5 + 32, %6
%endmacro

; PROCESS_128x2x4 first, off_{first,second}_{src,ref}, advance_at_end
%macro PROCESS_128x2x4 5-6 0
  PROCESS_64x2x4 %1, %2, %3, %2 + 64, %3 + 64
  PROCESS_64x2x4  0, %4, %5, %4 + 64, %5 + 64, %6
%endmacro

; void aom_sadNxNx4d_sse2(uint8_t *src,    int src_stride,
;                         uint8_t *ref[4], int ref_stride,
;                         uint32_t res[4]);
; where NxN = 64x64, 32x32, 16x16, 16x8, 8x16, 8x8, 8x4, 4x8 and 4x4
%macro SADNXN4D 2
%if UNIX64
cglobal sad%1x%2x4d, 5, 8, 8, src, src_stride, ref1, ref_stride, \
                              res, ref2, ref3, ref4
%else
cglobal sad%1x%2x4d, 4, 7, 8, src, src_stride, ref1, ref_stride, \
                              ref2, ref3, ref4
%endif
  movsxdifnidn src_strideq, src_strided
  movsxdifnidn ref_strideq, ref_strided
  mov                ref2q, [ref1q+gprsize*1]
  mov                ref3q, [ref1q+gprsize*2]
  mov                ref4q, [ref1q+gprsize*3]
  mov                ref1q, [ref1q+gprsize*0]

  PROCESS_%1x2x4 1, 0, 0, src_strideq, ref_strideq, 1
%rep (%2-4)/2
  PROCESS_%1x2x4 0, 0, 0, src_strideq, ref_strideq, 1
%endrep
  PROCESS_%1x2x4 0, 0, 0, src_strideq, ref_strideq, 0

%if %1 > 4
  pslldq                m5, 4
  pslldq                m7, 4
  por                   m4, m5
  por                   m6, m7
  mova                  m5, m4
  mova                  m7, m6
  punpcklqdq            m4, m6
  punpckhqdq            m5, m7
  movifnidn             r4, r4mp
  paddd                 m4, m5
  movu                [r4], m4
  RET
%else
  movifnidn             r4, r4mp
  pshufd            m6, m6, 0x08
  pshufd            m7, m7, 0x08
  movq              [r4+0], m6
  movq              [r4+8], m7
  RET
%endif
%endmacro

INIT_XMM sse2
SADNXN4D 128, 128
SADNXN4D 128, 64
SADNXN4D 64,  128
SADNXN4D 64, 64
SADNXN4D 64, 32
SADNXN4D 32, 64
SADNXN4D 32, 32
SADNXN4D 32, 16
SADNXN4D 16, 32
SADNXN4D 16, 16
SADNXN4D 16,  8
SADNXN4D  8, 16
SADNXN4D  8,  8
SADNXN4D  8,  4
SADNXN4D  4,  8
SADNXN4D  4,  4
SADNXN4D  4, 16
SADNXN4D 16,  4
SADNXN4D  8, 32
SADNXN4D 32,  8
SADNXN4D 16, 64
SADNXN4D 64, 16