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
|
/*
* Simd Library (http://ermig1979.github.io/Simd).
*
* Copyright (c) 2011-2021 Yermalayeu Ihar,
* 2014-2015 Antonenka Mikhail.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef __SimdConversion_h__
#define __SimdConversion_h__
#include "Simd/SimdConst.h"
#include "Simd/SimdMath.h"
#include "Simd/SimdLoad.h"
namespace Simd
{
namespace Base
{
SIMD_INLINE int BgrToGray(int blue, int green, int red)
{
return (BLUE_TO_GRAY_WEIGHT*blue + GREEN_TO_GRAY_WEIGHT * green +
RED_TO_GRAY_WEIGHT * red + BGR_TO_GRAY_ROUND_TERM) >> BGR_TO_GRAY_AVERAGING_SHIFT;
}
}
#ifdef SIMD_SSE41_ENABLE
namespace Sse41
{
template <int index> __m128i InterleaveBgr(__m128i blue, __m128i green, __m128i red);
template<> SIMD_INLINE __m128i InterleaveBgr<0>(__m128i blue, __m128i green, __m128i red)
{
return
_mm_or_si128(_mm_shuffle_epi8(blue, K8_SHUFFLE_BLUE_TO_BGR0),
_mm_or_si128(_mm_shuffle_epi8(green, K8_SHUFFLE_GREEN_TO_BGR0),
_mm_shuffle_epi8(red, K8_SHUFFLE_RED_TO_BGR0)));
}
template<> SIMD_INLINE __m128i InterleaveBgr<1>(__m128i blue, __m128i green, __m128i red)
{
return
_mm_or_si128(_mm_shuffle_epi8(blue, K8_SHUFFLE_BLUE_TO_BGR1),
_mm_or_si128(_mm_shuffle_epi8(green, K8_SHUFFLE_GREEN_TO_BGR1),
_mm_shuffle_epi8(red, K8_SHUFFLE_RED_TO_BGR1)));
}
template<> SIMD_INLINE __m128i InterleaveBgr<2>(__m128i blue, __m128i green, __m128i red)
{
return
_mm_or_si128(_mm_shuffle_epi8(blue, K8_SHUFFLE_BLUE_TO_BGR2),
_mm_or_si128(_mm_shuffle_epi8(green, K8_SHUFFLE_GREEN_TO_BGR2),
_mm_shuffle_epi8(red, K8_SHUFFLE_RED_TO_BGR2)));
}
SIMD_INLINE __m128i BgrToBlue(__m128i bgr[3])
{
return
_mm_or_si128(_mm_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_BLUE),
_mm_or_si128(_mm_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_BLUE),
_mm_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_BLUE)));
}
SIMD_INLINE __m128i BgrToGreen(__m128i bgr[3])
{
return
_mm_or_si128(_mm_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_GREEN),
_mm_or_si128(_mm_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_GREEN),
_mm_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_GREEN)));
}
SIMD_INLINE __m128i BgrToRed(__m128i bgr[3])
{
return
_mm_or_si128(_mm_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_RED),
_mm_or_si128(_mm_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_RED),
_mm_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_RED)));
}
}
#endif
#ifdef SIMD_AVX2_ENABLE
namespace Avx2
{
template <int index> __m256i GrayToBgr(__m256i gray);
template<> SIMD_INLINE __m256i GrayToBgr<0>(__m256i gray)
{
return _mm256_shuffle_epi8(_mm256_permute4x64_epi64(gray, 0x44), K8_SHUFFLE_GRAY_TO_BGR0);
}
template<> SIMD_INLINE __m256i GrayToBgr<1>(__m256i gray)
{
return _mm256_shuffle_epi8(_mm256_permute4x64_epi64(gray, 0x99), K8_SHUFFLE_GRAY_TO_BGR1);
}
template<> SIMD_INLINE __m256i GrayToBgr<2>(__m256i gray)
{
return _mm256_shuffle_epi8(_mm256_permute4x64_epi64(gray, 0xEE), K8_SHUFFLE_GRAY_TO_BGR2);
}
template <int index> __m256i InterleaveBgr(__m256i blue, __m256i green, __m256i red);
template<> SIMD_INLINE __m256i InterleaveBgr<0>(__m256i blue, __m256i green, __m256i red)
{
return
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(blue, 0x44), K8_SHUFFLE_PERMUTED_BLUE_TO_BGR0),
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(green, 0x44), K8_SHUFFLE_PERMUTED_GREEN_TO_BGR0),
_mm256_shuffle_epi8(_mm256_permute4x64_epi64(red, 0x44), K8_SHUFFLE_PERMUTED_RED_TO_BGR0)));
}
template<> SIMD_INLINE __m256i InterleaveBgr<1>(__m256i blue, __m256i green, __m256i red)
{
return
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(blue, 0x99), K8_SHUFFLE_PERMUTED_BLUE_TO_BGR1),
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(green, 0x99), K8_SHUFFLE_PERMUTED_GREEN_TO_BGR1),
_mm256_shuffle_epi8(_mm256_permute4x64_epi64(red, 0x99), K8_SHUFFLE_PERMUTED_RED_TO_BGR1)));
}
template<> SIMD_INLINE __m256i InterleaveBgr<2>(__m256i blue, __m256i green, __m256i red)
{
return
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(blue, 0xEE), K8_SHUFFLE_PERMUTED_BLUE_TO_BGR2),
_mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(green, 0xEE), K8_SHUFFLE_PERMUTED_GREEN_TO_BGR2),
_mm256_shuffle_epi8(_mm256_permute4x64_epi64(red, 0xEE), K8_SHUFFLE_PERMUTED_RED_TO_BGR2)));
}
SIMD_INLINE __m256i BgrToBlue(__m256i bgr[3])
{
__m256i b0 = _mm256_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_BLUE);
__m256i b2 = _mm256_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_BLUE);
return
_mm256_or_si256(_mm256_permute2x128_si256(b0, b2, 0x20),
_mm256_or_si256(_mm256_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_BLUE),
_mm256_permute2x128_si256(b0, b2, 0x31)));
}
SIMD_INLINE __m256i BgrToGreen(__m256i bgr[3])
{
__m256i g0 = _mm256_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_GREEN);
__m256i g2 = _mm256_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_GREEN);
return
_mm256_or_si256(_mm256_permute2x128_si256(g0, g2, 0x20),
_mm256_or_si256(_mm256_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_GREEN),
_mm256_permute2x128_si256(g0, g2, 0x31)));
}
SIMD_INLINE __m256i BgrToRed(__m256i bgr[3])
{
__m256i r0 = _mm256_shuffle_epi8(bgr[0], K8_SHUFFLE_BGR0_TO_RED);
__m256i r2 = _mm256_shuffle_epi8(bgr[2], K8_SHUFFLE_BGR2_TO_RED);
return
_mm256_or_si256(_mm256_permute2x128_si256(r0, r2, 0x20),
_mm256_or_si256(_mm256_shuffle_epi8(bgr[1], K8_SHUFFLE_BGR1_TO_RED),
_mm256_permute2x128_si256(r0, r2, 0x31)));
}
template<bool tail> __m256i BgrToBgra(const __m256i & bgr, const __m256i & alpha);
template<> SIMD_INLINE __m256i BgrToBgra<false>(const __m256i & bgr, const __m256i & alpha)
{
return _mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(bgr, 0x94), K8_BGR_TO_BGRA_SHUFFLE), alpha);
}
template<> SIMD_INLINE __m256i BgrToBgra<true>(const __m256i & bgr, const __m256i & alpha)
{
return _mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(bgr, 0xE9), K8_BGR_TO_BGRA_SHUFFLE), alpha);
}
template<bool tail> __m256i RgbToBgra(const __m256i & rgb, const __m256i & alpha);
template<> SIMD_INLINE __m256i RgbToBgra<false>(const __m256i & rgb, const __m256i & alpha)
{
return _mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(rgb, 0x94), K8_RGB_TO_BGRA_SHUFFLE), alpha);
}
template<> SIMD_INLINE __m256i RgbToBgra<true>(const __m256i & rgb, const __m256i & alpha)
{
return _mm256_or_si256(_mm256_shuffle_epi8(_mm256_permute4x64_epi64(rgb, 0xE9), K8_RGB_TO_BGRA_SHUFFLE), alpha);
}
}
#endif// SIMD_AVX2_ENABLE
#ifdef SIMD_NEON_ENABLE
namespace Neon
{
template <int part> SIMD_INLINE uint32x4_t BgrToGray(const uint16x8_t & blue, const uint16x8_t & green, const uint16x8_t & red)
{
return vshrq_n_u32(vmlal_u16(vmlal_u16(vmlal_u16(K32_BGR_TO_GRAY_ROUND_TERM, Half<part>(blue), K16_BLUE_TO_GRAY_WEIGHT),
Half<part>(green), K16_GREEN_TO_GRAY_WEIGHT), Half<part>(red), K16_RED_TO_GRAY_WEIGHT), Base::BGR_TO_GRAY_AVERAGING_SHIFT);
}
SIMD_INLINE uint16x8_t BgrToGray(const uint16x8_t & blue, const uint16x8_t & green, const uint16x8_t & red)
{
return PackU32(BgrToGray<0>(blue, green, red), BgrToGray<1>(blue, green, red));
}
template <int part> SIMD_INLINE int32x4_t BgrToU(uint16x8_t blue, uint16x8_t green, uint16x8_t red)
{
return vshrq_n_s32(vmlal_s16(vmlal_s16(vmlal_s16(K32_BGR_TO_YUV_ROUND_TERM, vreinterpret_s16_u16(Half<part>(blue)), K16_BLUE_TO_U_WEIGHT),
vreinterpret_s16_u16(Half<part>(green)), K16_GREEN_TO_U_WEIGHT), vreinterpret_s16_u16(Half<part>(red)), K16_RED_TO_U_WEIGHT), Base::BGR_TO_YUV_AVERAGING_SHIFT);
}
SIMD_INLINE int16x8_t BgrToU(uint16x8_t blue, uint16x8_t green, uint16x8_t red)
{
return vaddq_s16(K16_UV_ADJUST, PackI32(BgrToU<0>(blue, green, red), BgrToU<1>(blue, green, red)));
}
SIMD_INLINE uint8x16_t BgrToU(uint8x16_t blue, uint8x16_t green, uint8x16_t red)
{
return PackSaturatedI16(
BgrToU(UnpackU8<0>(blue), UnpackU8<0>(green), UnpackU8<0>(red)),
BgrToU(UnpackU8<1>(blue), UnpackU8<1>(green), UnpackU8<1>(red)));
}
}
#endif// SIMD_NEON_ENABLE
}
#endif//__SimdConversion_h__
|
