summaryrefslogtreecommitdiffstats
path: root/media/libaom/src/aom_dsp/x86/aom_convolve_copy_sse2.asm
blob: 7283c32b8cbdbf58945a17b2d5f1e79d3d9d2836 (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
;
; 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

%macro convolve_fn 1-2
%ifidn %1, avg
%define AUX_XMM_REGS 4
%else
%define AUX_XMM_REGS 0
%endif
%ifidn %2, highbd
%define pavg pavgw
cglobal %2_convolve_%1, 4, 7, 4+AUX_XMM_REGS, src, src_stride, \
                                              dst, dst_stride, \
                                              fx, fxs, fy, fys, w, h, bd
%else
%define pavg pavgb
cglobal convolve_%1, 4, 7, 4+AUX_XMM_REGS, src, src_stride, \
                                           dst, dst_stride, \
                                           fx, fxs, fy, fys, w, h
%endif
  mov r4d, dword wm
%ifidn %2, highbd
  shl r4d, 1
  shl srcq, 1
  shl src_strideq, 1
  shl dstq, 1
  shl dst_strideq, 1
%else
  cmp r4d, 4
  je .w4
%endif
  cmp r4d, 8
  je .w8
  cmp r4d, 16
  je .w16
  cmp r4d, 32
  je .w32

  cmp r4d, 64
  je .w64
%ifidn %2, highbd
  cmp r4d, 128
  je .w128

.w256:
  mov                    r4d, dword hm
.loop256:
  movu                    m0, [srcq]
  movu                    m1, [srcq+16]
  movu                    m2, [srcq+32]
  movu                    m3, [srcq+48]
%ifidn %1, avg
  pavg                    m0, [dstq]
  pavg                    m1, [dstq+16]
  pavg                    m2, [dstq+32]
  pavg                    m3, [dstq+48]
%endif
  mova             [dstq   ], m0
  mova             [dstq+16], m1
  mova             [dstq+32], m2
  mova             [dstq+48], m3
  movu                    m0, [srcq+64]
  movu                    m1, [srcq+80]
  movu                    m2, [srcq+96]
  movu                    m3, [srcq+112]
%ifidn %1, avg
  pavg                    m0, [dstq+64]
  pavg                    m1, [dstq+80]
  pavg                    m2, [dstq+96]
  pavg                    m3, [dstq+112]
%endif
  mova             [dstq+64], m0
  mova             [dstq+80], m1
  mova             [dstq+96], m2
  mova            [dstq+112], m3
  movu                    m0, [srcq+128]
  movu                    m1, [srcq+128+16]
  movu                    m2, [srcq+128+32]
  movu                    m3, [srcq+128+48]
%ifidn %1, avg
  pavg                    m0, [dstq+128]
  pavg                    m1, [dstq+128+16]
  pavg                    m2, [dstq+128+32]
  pavg                    m3, [dstq+128+48]
%endif
  mova         [dstq+128   ], m0
  mova         [dstq+128+16], m1
  mova         [dstq+128+32], m2
  mova         [dstq+128+48], m3
  movu                    m0, [srcq+128+64]
  movu                    m1, [srcq+128+80]
  movu                    m2, [srcq+128+96]
  movu                    m3, [srcq+128+112]
  add                   srcq, src_strideq
%ifidn %1, avg
  pavg                    m0, [dstq+128+64]
  pavg                    m1, [dstq+128+80]
  pavg                    m2, [dstq+128+96]
  pavg                    m3, [dstq+128+112]
%endif
  mova         [dstq+128+64], m0
  mova         [dstq+128+80], m1
  mova         [dstq+128+96], m2
  mova        [dstq+128+112], m3
  add                   dstq, dst_strideq
  sub                    r4d, 1
  jnz .loop256
  RET
%endif

.w128:
  mov                    r4d, dword hm
.loop128:
  movu                    m0, [srcq]
  movu                    m1, [srcq+16]
  movu                    m2, [srcq+32]
  movu                    m3, [srcq+48]
%ifidn %1, avg
  pavg                    m0, [dstq]
  pavg                    m1, [dstq+16]
  pavg                    m2, [dstq+32]
  pavg                    m3, [dstq+48]
%endif
  mova             [dstq   ], m0
  mova             [dstq+16], m1
  mova             [dstq+32], m2
  mova             [dstq+48], m3
  movu                    m0, [srcq+64]
  movu                    m1, [srcq+80]
  movu                    m2, [srcq+96]
  movu                    m3, [srcq+112]
  add                   srcq, src_strideq
%ifidn %1, avg
  pavg                    m0, [dstq+64]
  pavg                    m1, [dstq+80]
  pavg                    m2, [dstq+96]
  pavg                    m3, [dstq+112]
%endif
  mova             [dstq+64], m0
  mova             [dstq+80], m1
  mova             [dstq+96], m2
  mova            [dstq+112], m3
  add                   dstq, dst_strideq
  sub                    r4d, 1
  jnz .loop128
  RET

.w64:
  mov                    r4d, dword hm
.loop64:
  movu                    m0, [srcq]
  movu                    m1, [srcq+16]
  movu                    m2, [srcq+32]
  movu                    m3, [srcq+48]
  add                   srcq, src_strideq
%ifidn %1, avg
  pavg                    m0, [dstq]
  pavg                    m1, [dstq+16]
  pavg                    m2, [dstq+32]
  pavg                    m3, [dstq+48]
%endif
  mova             [dstq   ], m0
  mova             [dstq+16], m1
  mova             [dstq+32], m2
  mova             [dstq+48], m3
  add                   dstq, dst_strideq
  sub                    r4d, 1
  jnz .loop64
  RET

.w32:
  mov                    r4d, dword hm
.loop32:
  movu                    m0, [srcq]
  movu                    m1, [srcq+16]
  movu                    m2, [srcq+src_strideq]
  movu                    m3, [srcq+src_strideq+16]
  lea                   srcq, [srcq+src_strideq*2]
%ifidn %1, avg
  pavg                    m0, [dstq]
  pavg                    m1, [dstq            +16]
  pavg                    m2, [dstq+dst_strideq]
  pavg                    m3, [dstq+dst_strideq+16]
%endif
  mova [dstq               ], m0
  mova [dstq            +16], m1
  mova [dstq+dst_strideq   ], m2
  mova [dstq+dst_strideq+16], m3
  lea                   dstq, [dstq+dst_strideq*2]
  sub                    r4d, 2
  jnz .loop32
  RET

.w16:
  mov                    r4d, dword hm
  lea                    r5q, [src_strideq*3]
  lea                    r6q, [dst_strideq*3]
.loop16:
  movu                    m0, [srcq]
  movu                    m1, [srcq+src_strideq]
  movu                    m2, [srcq+src_strideq*2]
  movu                    m3, [srcq+r5q]
  lea                   srcq, [srcq+src_strideq*4]
%ifidn %1, avg
  pavg                    m0, [dstq]
  pavg                    m1, [dstq+dst_strideq]
  pavg                    m2, [dstq+dst_strideq*2]
  pavg                    m3, [dstq+r6q]
%endif
  mova  [dstq              ], m0
  mova  [dstq+dst_strideq  ], m1
  mova  [dstq+dst_strideq*2], m2
  mova  [dstq+r6q          ], m3
  lea                   dstq, [dstq+dst_strideq*4]
  sub                    r4d, 4
  jnz .loop16
  RET

.w8:
  mov                    r4d, dword hm
  lea                    r5q, [src_strideq*3]
  lea                    r6q, [dst_strideq*3]
.loop8:
  movh                    m0, [srcq]
  movh                    m1, [srcq+src_strideq]
  movh                    m2, [srcq+src_strideq*2]
  movh                    m3, [srcq+r5q]
  lea                   srcq, [srcq+src_strideq*4]
%ifidn %1, avg
  movh                    m4, [dstq]
  movh                    m5, [dstq+dst_strideq]
  movh                    m6, [dstq+dst_strideq*2]
  movh                    m7, [dstq+r6q]
  pavg                    m0, m4
  pavg                    m1, m5
  pavg                    m2, m6
  pavg                    m3, m7
%endif
  movh  [dstq              ], m0
  movh  [dstq+dst_strideq  ], m1
  movh  [dstq+dst_strideq*2], m2
  movh  [dstq+r6q          ], m3
  lea                   dstq, [dstq+dst_strideq*4]
  sub                    r4d, 4
  jnz .loop8
  RET

%ifnidn %2, highbd
.w4:
  mov                    r4d, dword hm
  lea                    r5q, [src_strideq*3]
  lea                    r6q, [dst_strideq*3]
.loop4:
  movd                    m0, [srcq]
  movd                    m1, [srcq+src_strideq]
  movd                    m2, [srcq+src_strideq*2]
  movd                    m3, [srcq+r5q]
  lea                   srcq, [srcq+src_strideq*4]
%ifidn %1, avg
  movd                    m4, [dstq]
  movd                    m5, [dstq+dst_strideq]
  movd                    m6, [dstq+dst_strideq*2]
  movd                    m7, [dstq+r6q]
  pavg                    m0, m4
  pavg                    m1, m5
  pavg                    m2, m6
  pavg                    m3, m7
%endif
  movd  [dstq              ], m0
  movd  [dstq+dst_strideq  ], m1
  movd  [dstq+dst_strideq*2], m2
  movd  [dstq+r6q          ], m3
  lea                   dstq, [dstq+dst_strideq*4]
  sub                    r4d, 4
  jnz .loop4
  RET
%endif
%endmacro

INIT_XMM sse2
convolve_fn copy
convolve_fn avg
convolve_fn copy, highbd