Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(346)

Issues Search
    My Issues | Starred     Open | Closed | All

Side by Side Diff: source/libvpx/third_party/libyuv/source/row_any.cc

Issue 1302353004: libvpx: Pull from upstream (Closed) Base URL: https://chromium.googlesource.com/chromium/deps/libvpx.git@master
Patch Set: Created 5 years, 3 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
« no previous file with comments | « source/libvpx/third_party/libyuv/source/rotate_win.cc ('k') | source/libvpx/third_party/libyuv/source/row_common.cc » ('j') | no next file with comments »
Toggle Intra-line Diffs ('i') | Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
OLDNEW
1 /* 1 /*
2 * Copyright 2012 The LibYuv Project Authors. All rights reserved. 2 * Copyright 2012 The LibYuv Project Authors. All rights reserved.
3 * 3 *
4 * Use of this source code is governed by a BSD-style license 4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source 5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found 6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may 7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree. 8 * be found in the AUTHORS file in the root of the source tree.
9 */ 9 */
10 10
11 #include "libyuv/row.h" 11 #include "libyuv/row.h"
12 12
13 #include <string.h> // For memset.
14
13 #include "libyuv/basic_types.h" 15 #include "libyuv/basic_types.h"
14 16
15 #ifdef __cplusplus 17 #ifdef __cplusplus
16 namespace libyuv { 18 namespace libyuv {
17 extern "C" { 19 extern "C" {
18 #endif 20 #endif
19 21
20 // YUV to RGB does multiple of 8 with SIMD and remainder with C. 22 // Subsampled source needs to be increase by 1 of not even.
21 #define YANY(NAMEANY, I420TORGB_SIMD, I420TORGB_C, UV_SHIFT, BPP, MASK) \ 23 #define SS(width, shift) (((width) + (1 << (shift)) - 1) >> (shift))
24
25 // Any 3 planes to 1.
26 #define ANY31(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \
22 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \ 27 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \
23 uint8* rgb_buf, int width) { \ 28 uint8* dst_ptr, int width) { \
24 int n = width & ~MASK; \ 29 SIMD_ALIGNED(uint8 temp[64 * 4]); \
25 if (n > 0) { \ 30 memset(temp, 0, 64 * 3); /* for YUY2 and msan */ \
26 I420TORGB_SIMD(y_buf, u_buf, v_buf, rgb_buf, n); \ 31 int r = width & MASK; \
27 } \ 32 int n = width & ~MASK; \
28 I420TORGB_C(y_buf + n, \ 33 if (n > 0) { \
29 u_buf + (n >> UV_SHIFT), \ 34 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, n); \
30 v_buf + (n >> UV_SHIFT), \ 35 } \
31 rgb_buf + n * BPP, width & MASK); \ 36 memcpy(temp, y_buf + n, r); \
37 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \
38 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \
39 ANY_SIMD(temp, temp + 64, temp + 128, temp + 192, MASK + 1); \
40 memcpy(dst_ptr + (n >> DUVSHIFT) * BPP, temp + 192, \
41 SS(r, DUVSHIFT) * BPP); \
32 } 42 }
33 43
34 #ifdef HAS_I422TOARGBROW_SSSE3 44 #ifdef HAS_I422TOARGBROW_SSSE3
35 YANY(I422ToARGBRow_Any_SSSE3, I422ToARGBRow_SSSE3, I422ToARGBRow_C, 45 ANY31(I422ToARGBRow_Any_SSSE3, I422ToARGBRow_SSSE3, 1, 0, 4, 7)
36 1, 4, 7)
37 #endif 46 #endif
38 #ifdef HAS_I444TOARGBROW_SSSE3 47 #ifdef HAS_I444TOARGBROW_SSSE3
39 YANY(I444ToARGBRow_Any_SSSE3, I444ToARGBRow_SSSE3, I444ToARGBRow_C, 48 ANY31(I444ToARGBRow_Any_SSSE3, I444ToARGBRow_SSSE3, 0, 0, 4, 7)
40 0, 4, 7) 49 ANY31(I411ToARGBRow_Any_SSSE3, I411ToARGBRow_SSSE3, 2, 0, 4, 7)
41 YANY(I411ToARGBRow_Any_SSSE3, I411ToARGBRow_SSSE3, I411ToARGBRow_C, 50 ANY31(I422ToBGRARow_Any_SSSE3, I422ToBGRARow_SSSE3, 1, 0, 4, 7)
42 2, 4, 7) 51 ANY31(I422ToABGRRow_Any_SSSE3, I422ToABGRRow_SSSE3, 1, 0, 4, 7)
43 YANY(I422ToBGRARow_Any_SSSE3, I422ToBGRARow_SSSE3, I422ToBGRARow_C, 52 ANY31(I422ToRGBARow_Any_SSSE3, I422ToRGBARow_SSSE3, 1, 0, 4, 7)
44 1, 4, 7) 53 ANY31(I422ToARGB4444Row_Any_SSSE3, I422ToARGB4444Row_SSSE3, 1, 0, 2, 7)
45 YANY(I422ToABGRRow_Any_SSSE3, I422ToABGRRow_SSSE3, I422ToABGRRow_C, 54 ANY31(I422ToARGB1555Row_Any_SSSE3, I422ToARGB1555Row_SSSE3, 1, 0, 2, 7)
46 1, 4, 7) 55 ANY31(I422ToRGB565Row_Any_SSSE3, I422ToRGB565Row_SSSE3, 1, 0, 2, 7)
47 YANY(I422ToRGBARow_Any_SSSE3, I422ToRGBARow_SSSE3, I422ToRGBARow_C, 56 ANY31(I422ToRGB24Row_Any_SSSE3, I422ToRGB24Row_SSSE3, 1, 0, 3, 7)
48 1, 4, 7) 57 ANY31(I422ToRAWRow_Any_SSSE3, I422ToRAWRow_SSSE3, 1, 0, 3, 7)
49 YANY(I422ToARGB4444Row_Any_SSSE3, I422ToARGB4444Row_SSSE3, I422ToARGB4444Row_C, 58 ANY31(I422ToYUY2Row_Any_SSE2, I422ToYUY2Row_SSE2, 1, 1, 4, 15)
50 1, 2, 7) 59 ANY31(I422ToUYVYRow_Any_SSE2, I422ToUYVYRow_SSE2, 1, 1, 4, 15)
51 YANY(I422ToARGB1555Row_Any_SSSE3, I422ToARGB1555Row_SSSE3, I422ToARGB1555Row_C,
52 1, 2, 7)
53 YANY(I422ToRGB565Row_Any_SSSE3, I422ToRGB565Row_SSSE3, I422ToRGB565Row_C,
54 1, 2, 7)
55 YANY(I422ToRGB24Row_Any_SSSE3, I422ToRGB24Row_SSSE3, I422ToRGB24Row_C, 1, 3, 7)
56 YANY(I422ToRAWRow_Any_SSSE3, I422ToRAWRow_SSSE3, I422ToRAWRow_C, 1, 3, 7)
57 YANY(I422ToYUY2Row_Any_SSE2, I422ToYUY2Row_SSE2, I422ToYUY2Row_C, 1, 2, 15)
58 YANY(I422ToUYVYRow_Any_SSE2, I422ToUYVYRow_SSE2, I422ToUYVYRow_C, 1, 2, 15)
59 #endif // HAS_I444TOARGBROW_SSSE3 60 #endif // HAS_I444TOARGBROW_SSSE3
61 #ifdef HAS_I422TORGB24ROW_AVX2
62 ANY31(I422ToRGB24Row_Any_AVX2, I422ToRGB24Row_AVX2, 1, 0, 3, 15)
63 #endif
64 #ifdef HAS_I422TORAWROW_AVX2
65 ANY31(I422ToRAWRow_Any_AVX2, I422ToRAWRow_AVX2, 1, 0, 3, 15)
66 #endif
60 #ifdef HAS_J422TOARGBROW_SSSE3 67 #ifdef HAS_J422TOARGBROW_SSSE3
61 YANY(J422ToARGBRow_Any_SSSE3, J422ToARGBRow_SSSE3, J422ToARGBRow_C, 68 ANY31(J422ToARGBRow_Any_SSSE3, J422ToARGBRow_SSSE3, 1, 0, 4, 7)
62 1, 4, 7) 69 #endif
70 #ifdef HAS_J422TOARGBROW_AVX2
71 ANY31(J422ToARGBRow_Any_AVX2, J422ToARGBRow_AVX2, 1, 0, 4, 15)
63 #endif 72 #endif
64 #ifdef HAS_I422TOARGBROW_AVX2 73 #ifdef HAS_I422TOARGBROW_AVX2
65 YANY(I422ToARGBRow_Any_AVX2, I422ToARGBRow_AVX2, I422ToARGBRow_C, 1, 4, 15) 74 ANY31(I422ToARGBRow_Any_AVX2, I422ToARGBRow_AVX2, 1, 0, 4, 15)
66 #endif 75 #endif
67 #ifdef HAS_I422TOBGRAROW_AVX2 76 #ifdef HAS_I422TOBGRAROW_AVX2
68 YANY(I422ToBGRARow_Any_AVX2, I422ToBGRARow_AVX2, I422ToBGRARow_C, 1, 4, 15) 77 ANY31(I422ToBGRARow_Any_AVX2, I422ToBGRARow_AVX2, 1, 0, 4, 15)
69 #endif 78 #endif
70 #ifdef HAS_I422TORGBAROW_AVX2 79 #ifdef HAS_I422TORGBAROW_AVX2
71 YANY(I422ToRGBARow_Any_AVX2, I422ToRGBARow_AVX2, I422ToRGBARow_C, 1, 4, 15) 80 ANY31(I422ToRGBARow_Any_AVX2, I422ToRGBARow_AVX2, 1, 0, 4, 15)
72 #endif 81 #endif
73 #ifdef HAS_I422TOABGRROW_AVX2 82 #ifdef HAS_I422TOABGRROW_AVX2
74 YANY(I422ToABGRRow_Any_AVX2, I422ToABGRRow_AVX2, I422ToABGRRow_C, 1, 4, 15) 83 ANY31(I422ToABGRRow_Any_AVX2, I422ToABGRRow_AVX2, 1, 0, 4, 15)
84 #endif
85 #ifdef HAS_I444TOARGBROW_AVX2
86 ANY31(I444ToARGBRow_Any_AVX2, I444ToARGBRow_AVX2, 0, 0, 4, 15)
87 #endif
88 #ifdef HAS_I411TOARGBROW_AVX2
89 ANY31(I411ToARGBRow_Any_AVX2, I411ToARGBRow_AVX2, 2, 0, 4, 15)
75 #endif 90 #endif
76 #ifdef HAS_I422TOARGB4444ROW_AVX2 91 #ifdef HAS_I422TOARGB4444ROW_AVX2
77 YANY(I422ToARGB4444Row_Any_AVX2, I422ToARGB4444Row_AVX2, I422ToARGB4444Row_C, 92 ANY31(I422ToARGB4444Row_Any_AVX2, I422ToARGB4444Row_AVX2, 1, 0, 2, 7)
78 1, 2, 7)
79 #endif 93 #endif
80 #ifdef HAS_I422TOARGB1555ROW_AVX2 94 #ifdef HAS_I422TOARGB1555ROW_AVX2
81 YANY(I422ToARGB1555Row_Any_AVX2, I422ToARGB1555Row_AVX2, I422ToARGB1555Row_C, 95 ANY31(I422ToARGB1555Row_Any_AVX2, I422ToARGB1555Row_AVX2, 1, 0, 2, 7)
82 1, 2, 7)
83 #endif 96 #endif
84 #ifdef HAS_I422TORGB565ROW_AVX2 97 #ifdef HAS_I422TORGB565ROW_AVX2
85 YANY(I422ToRGB565Row_Any_AVX2, I422ToRGB565Row_AVX2, I422ToRGB565Row_C, 98 ANY31(I422ToRGB565Row_Any_AVX2, I422ToRGB565Row_AVX2, 1, 0, 2, 7)
86 1, 2, 7)
87 #endif 99 #endif
88 #ifdef HAS_I422TOARGBROW_NEON 100 #ifdef HAS_I422TOARGBROW_NEON
89 YANY(I444ToARGBRow_Any_NEON, I444ToARGBRow_NEON, I444ToARGBRow_C, 0, 4, 7) 101 ANY31(I444ToARGBRow_Any_NEON, I444ToARGBRow_NEON, 0, 0, 4, 7)
90 YANY(I422ToARGBRow_Any_NEON, I422ToARGBRow_NEON, I422ToARGBRow_C, 1, 4, 7) 102 ANY31(I422ToARGBRow_Any_NEON, I422ToARGBRow_NEON, 1, 0, 4, 7)
91 YANY(I411ToARGBRow_Any_NEON, I411ToARGBRow_NEON, I411ToARGBRow_C, 2, 4, 7) 103 ANY31(I411ToARGBRow_Any_NEON, I411ToARGBRow_NEON, 2, 0, 4, 7)
92 YANY(I422ToBGRARow_Any_NEON, I422ToBGRARow_NEON, I422ToBGRARow_C, 1, 4, 7) 104 ANY31(I422ToBGRARow_Any_NEON, I422ToBGRARow_NEON, 1, 0, 4, 7)
93 YANY(I422ToABGRRow_Any_NEON, I422ToABGRRow_NEON, I422ToABGRRow_C, 1, 4, 7) 105 ANY31(I422ToABGRRow_Any_NEON, I422ToABGRRow_NEON, 1, 0, 4, 7)
94 YANY(I422ToRGBARow_Any_NEON, I422ToRGBARow_NEON, I422ToRGBARow_C, 1, 4, 7) 106 ANY31(I422ToRGBARow_Any_NEON, I422ToRGBARow_NEON, 1, 0, 4, 7)
95 YANY(I422ToRGB24Row_Any_NEON, I422ToRGB24Row_NEON, I422ToRGB24Row_C, 1, 3, 7) 107 ANY31(I422ToRGB24Row_Any_NEON, I422ToRGB24Row_NEON, 1, 0, 3, 7)
96 YANY(I422ToRAWRow_Any_NEON, I422ToRAWRow_NEON, I422ToRAWRow_C, 1, 3, 7) 108 ANY31(I422ToRAWRow_Any_NEON, I422ToRAWRow_NEON, 1, 0, 3, 7)
97 YANY(I422ToARGB4444Row_Any_NEON, I422ToARGB4444Row_NEON, I422ToARGB4444Row_C, 109 ANY31(I422ToARGB4444Row_Any_NEON, I422ToARGB4444Row_NEON, 1, 0, 2, 7)
98 1, 2, 7) 110 ANY31(I422ToARGB1555Row_Any_NEON, I422ToARGB1555Row_NEON, 1, 0, 2, 7)
99 YANY(I422ToARGB1555Row_Any_NEON, I422ToARGB1555Row_NEON, I422ToARGB1555Row_C, 111 ANY31(I422ToRGB565Row_Any_NEON, I422ToRGB565Row_NEON, 1, 0, 2, 7)
100 1, 2, 7)
101 YANY(I422ToRGB565Row_Any_NEON, I422ToRGB565Row_NEON, I422ToRGB565Row_C, 1, 2, 7)
102 #endif 112 #endif
103 #ifdef HAS_I422TOYUY2ROW_NEON 113 #ifdef HAS_I422TOYUY2ROW_NEON
104 YANY(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, I422ToYUY2Row_C, 1, 2, 15) 114 ANY31(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, 1, 1, 4, 15)
105 #endif 115 #endif
106 #ifdef HAS_I422TOUYVYROW_NEON 116 #ifdef HAS_I422TOUYVYROW_NEON
107 YANY(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, I422ToUYVYRow_C, 1, 2, 15) 117 ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15)
108 #endif 118 #endif
109 #undef YANY 119 #undef ANY31
110 120
111 // Wrappers to handle odd width 121 // Any 2 planes to 1.
112 #define NV2NY(NAMEANY, NV12TORGB_SIMD, NV12TORGB_C, UV_SHIFT, BPP, MASK) \ 122 #define ANY21(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, SBPP2, BPP, MASK) \
113 void NAMEANY(const uint8* y_buf, const uint8* uv_buf, \ 123 void NAMEANY(const uint8* y_buf, const uint8* uv_buf, \
114 uint8* rgb_buf, int width) { \ 124 uint8* dst_ptr, int width) { \
115 int n = width & ~MASK; \ 125 SIMD_ALIGNED(uint8 temp[64 * 3]); \
116 if (n > 0) { \ 126 memset(temp, 0, 64 * 2); /* for msan */ \
117 NV12TORGB_SIMD(y_buf, uv_buf, rgb_buf, n); \ 127 int r = width & MASK; \
118 } \ 128 int n = width & ~MASK; \
119 NV12TORGB_C(y_buf + n, \ 129 if (n > 0) { \
120 uv_buf + (n >> UV_SHIFT), \ 130 ANY_SIMD(y_buf, uv_buf, dst_ptr, n); \
121 rgb_buf + n * BPP, width & MASK); \ 131 } \
122 } 132 memcpy(temp, y_buf + n * SBPP, r * SBPP); \
123 133 memcpy(temp + 64, uv_buf + (n >> UVSHIFT) * SBPP2, \
134 SS(r, UVSHIFT) * SBPP2); \
135 ANY_SIMD(temp, temp + 64, temp + 128, MASK + 1); \
136 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \
137 }
138
139 // Biplanar to RGB.
124 #ifdef HAS_NV12TOARGBROW_SSSE3 140 #ifdef HAS_NV12TOARGBROW_SSSE3
125 NV2NY(NV12ToARGBRow_Any_SSSE3, NV12ToARGBRow_SSSE3, NV12ToARGBRow_C, 0, 4, 7) 141 ANY21(NV12ToARGBRow_Any_SSSE3, NV12ToARGBRow_SSSE3, 1, 1, 2, 4, 7)
126 NV2NY(NV21ToARGBRow_Any_SSSE3, NV21ToARGBRow_SSSE3, NV21ToARGBRow_C, 0, 4, 7) 142 ANY21(NV21ToARGBRow_Any_SSSE3, NV21ToARGBRow_SSSE3, 1, 1, 2, 4, 7)
127 #endif 143 #endif
128 #ifdef HAS_NV12TOARGBROW_AVX2 144 #ifdef HAS_NV12TOARGBROW_AVX2
129 NV2NY(NV12ToARGBRow_Any_AVX2, NV12ToARGBRow_AVX2, NV12ToARGBRow_C, 0, 4, 15) 145 ANY21(NV12ToARGBRow_Any_AVX2, NV12ToARGBRow_AVX2, 1, 1, 2, 4, 15)
130 NV2NY(NV21ToARGBRow_Any_AVX2, NV21ToARGBRow_AVX2, NV21ToARGBRow_C, 0, 4, 15) 146 ANY21(NV21ToARGBRow_Any_AVX2, NV21ToARGBRow_AVX2, 1, 1, 2, 4, 15)
131 #endif 147 #endif
132 #ifdef HAS_NV12TOARGBROW_NEON 148 #ifdef HAS_NV12TOARGBROW_NEON
133 NV2NY(NV12ToARGBRow_Any_NEON, NV12ToARGBRow_NEON, NV12ToARGBRow_C, 0, 4, 7) 149 ANY21(NV12ToARGBRow_Any_NEON, NV12ToARGBRow_NEON, 1, 1, 2, 4, 7)
134 NV2NY(NV21ToARGBRow_Any_NEON, NV21ToARGBRow_NEON, NV21ToARGBRow_C, 0, 4, 7) 150 ANY21(NV21ToARGBRow_Any_NEON, NV21ToARGBRow_NEON, 1, 1, 2, 4, 7)
135 #endif 151 #endif
136 #ifdef HAS_NV12TORGB565ROW_SSSE3 152 #ifdef HAS_NV12TORGB565ROW_SSSE3
137 NV2NY(NV12ToRGB565Row_Any_SSSE3, NV12ToRGB565Row_SSSE3, NV12ToRGB565Row_C, 153 ANY21(NV12ToRGB565Row_Any_SSSE3, NV12ToRGB565Row_SSSE3, 1, 1, 2, 2, 7)
138 0, 2, 7) 154 ANY21(NV21ToRGB565Row_Any_SSSE3, NV21ToRGB565Row_SSSE3, 1, 1, 2, 2, 7)
139 NV2NY(NV21ToRGB565Row_Any_SSSE3, NV21ToRGB565Row_SSSE3, NV21ToRGB565Row_C,
140 0, 2, 7)
141 #endif 155 #endif
142 #ifdef HAS_NV12TORGB565ROW_AVX2 156 #ifdef HAS_NV12TORGB565ROW_AVX2
143 NV2NY(NV12ToRGB565Row_Any_AVX2, NV12ToRGB565Row_AVX2, NV12ToRGB565Row_C, 157 ANY21(NV12ToRGB565Row_Any_AVX2, NV12ToRGB565Row_AVX2, 1, 1, 2, 2, 15)
144 0, 2, 15) 158 ANY21(NV21ToRGB565Row_Any_AVX2, NV21ToRGB565Row_AVX2, 1, 1, 2, 2, 15)
145 NV2NY(NV21ToRGB565Row_Any_AVX2, NV21ToRGB565Row_AVX2, NV21ToRGB565Row_C,
146 0, 2, 15)
147 #endif 159 #endif
148 #ifdef HAS_NV12TORGB565ROW_NEON 160 #ifdef HAS_NV12TORGB565ROW_NEON
149 NV2NY(NV12ToRGB565Row_Any_NEON, NV12ToRGB565Row_NEON, NV12ToRGB565Row_C, 161 ANY21(NV12ToRGB565Row_Any_NEON, NV12ToRGB565Row_NEON, 1, 1, 2, 2, 7)
150 0, 2, 7) 162 ANY21(NV21ToRGB565Row_Any_NEON, NV21ToRGB565Row_NEON, 1, 1, 2, 2, 7)
151 NV2NY(NV21ToRGB565Row_Any_NEON, NV21ToRGB565Row_NEON, NV21ToRGB565Row_C, 163 #endif
152 0, 2, 7) 164
153 #endif 165 // Merge functions.
154 #undef NVANY 166 #ifdef HAS_MERGEUVROW_SSE2
155 167 ANY21(MergeUVRow_Any_SSE2, MergeUVRow_SSE2, 0, 1, 1, 2, 15)
156 #define RGBANY(NAMEANY, ARGBTORGB_SIMD, ARGBTORGB_C, SBPP, BPP, MASK) \ 168 #endif
157 void NAMEANY(const uint8* src, uint8* dst, int width) { \ 169 #ifdef HAS_MERGEUVROW_AVX2
158 int n = width & ~MASK; \ 170 ANY21(MergeUVRow_Any_AVX2, MergeUVRow_AVX2, 0, 1, 1, 2, 31)
159 if (n > 0) { \ 171 #endif
160 ARGBTORGB_SIMD(src, dst, n); \ 172 #ifdef HAS_MERGEUVROW_NEON
161 } \ 173 ANY21(MergeUVRow_Any_NEON, MergeUVRow_NEON, 0, 1, 1, 2, 15)
162 ARGBTORGB_C(src + n * SBPP, dst + n * BPP, width & MASK); \ 174 #endif
163 } 175
164 176 // Math functions.
177 #ifdef HAS_ARGBMULTIPLYROW_SSE2
178 ANY21(ARGBMultiplyRow_Any_SSE2, ARGBMultiplyRow_SSE2, 0, 4, 4, 4, 3)
179 #endif
180 #ifdef HAS_ARGBADDROW_SSE2
181 ANY21(ARGBAddRow_Any_SSE2, ARGBAddRow_SSE2, 0, 4, 4, 4, 3)
182 #endif
183 #ifdef HAS_ARGBSUBTRACTROW_SSE2
184 ANY21(ARGBSubtractRow_Any_SSE2, ARGBSubtractRow_SSE2, 0, 4, 4, 4, 3)
185 #endif
186 #ifdef HAS_ARGBMULTIPLYROW_AVX2
187 ANY21(ARGBMultiplyRow_Any_AVX2, ARGBMultiplyRow_AVX2, 0, 4, 4, 4, 7)
188 #endif
189 #ifdef HAS_ARGBADDROW_AVX2
190 ANY21(ARGBAddRow_Any_AVX2, ARGBAddRow_AVX2, 0, 4, 4, 4, 7)
191 #endif
192 #ifdef HAS_ARGBSUBTRACTROW_AVX2
193 ANY21(ARGBSubtractRow_Any_AVX2, ARGBSubtractRow_AVX2, 0, 4, 4, 4, 7)
194 #endif
195 #ifdef HAS_ARGBMULTIPLYROW_NEON
196 ANY21(ARGBMultiplyRow_Any_NEON, ARGBMultiplyRow_NEON, 0, 4, 4, 4, 7)
197 #endif
198 #ifdef HAS_ARGBADDROW_NEON
199 ANY21(ARGBAddRow_Any_NEON, ARGBAddRow_NEON, 0, 4, 4, 4, 7)
200 #endif
201 #ifdef HAS_ARGBSUBTRACTROW_NEON
202 ANY21(ARGBSubtractRow_Any_NEON, ARGBSubtractRow_NEON, 0, 4, 4, 4, 7)
203 #endif
204 #ifdef HAS_SOBELROW_SSE2
205 ANY21(SobelRow_Any_SSE2, SobelRow_SSE2, 0, 1, 1, 4, 15)
206 #endif
207 #ifdef HAS_SOBELROW_NEON
208 ANY21(SobelRow_Any_NEON, SobelRow_NEON, 0, 1, 1, 4, 7)
209 #endif
210 #ifdef HAS_SOBELTOPLANEROW_SSE2
211 ANY21(SobelToPlaneRow_Any_SSE2, SobelToPlaneRow_SSE2, 0, 1, 1, 1, 15)
212 #endif
213 #ifdef HAS_SOBELTOPLANEROW_NEON
214 ANY21(SobelToPlaneRow_Any_NEON, SobelToPlaneRow_NEON, 0, 1, 1, 1, 15)
215 #endif
216 #ifdef HAS_SOBELXYROW_SSE2
217 ANY21(SobelXYRow_Any_SSE2, SobelXYRow_SSE2, 0, 1, 1, 4, 15)
218 #endif
219 #ifdef HAS_SOBELXYROW_NEON
220 ANY21(SobelXYRow_Any_NEON, SobelXYRow_NEON, 0, 1, 1, 4, 7)
221 #endif
222 #undef ANY21
223
224 // Any 1 to 1.
225 #define ANY11(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \
226 void NAMEANY(const uint8* src_ptr, uint8* dst_ptr, int width) { \
227 SIMD_ALIGNED(uint8 temp[128 * 2]); \
228 memset(temp, 0, 128); /* for YUY2 and msan */ \
229 int r = width & MASK; \
230 int n = width & ~MASK; \
231 if (n > 0) { \
232 ANY_SIMD(src_ptr, dst_ptr, n); \
233 } \
234 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \
235 ANY_SIMD(temp, temp + 128, MASK + 1); \
236 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \
237 }
238
239 #ifdef HAS_COPYROW_AVX
240 ANY11(CopyRow_Any_AVX, CopyRow_AVX, 0, 1, 1, 63)
241 #endif
242 #ifdef HAS_COPYROW_SSE2
243 ANY11(CopyRow_Any_SSE2, CopyRow_SSE2, 0, 1, 1, 31)
244 #endif
245 #ifdef HAS_COPYROW_NEON
246 ANY11(CopyRow_Any_NEON, CopyRow_NEON, 0, 1, 1, 31)
247 #endif
165 #if defined(HAS_ARGBTORGB24ROW_SSSE3) 248 #if defined(HAS_ARGBTORGB24ROW_SSSE3)
166 RGBANY(ARGBToRGB24Row_Any_SSSE3, ARGBToRGB24Row_SSSE3, ARGBToRGB24Row_C, 249 ANY11(ARGBToRGB24Row_Any_SSSE3, ARGBToRGB24Row_SSSE3, 0, 4, 3, 15)
167 4, 3, 15) 250 ANY11(ARGBToRAWRow_Any_SSSE3, ARGBToRAWRow_SSSE3, 0, 4, 3, 15)
168 RGBANY(ARGBToRAWRow_Any_SSSE3, ARGBToRAWRow_SSSE3, ARGBToRAWRow_C, 251 ANY11(ARGBToRGB565Row_Any_SSE2, ARGBToRGB565Row_SSE2, 0, 4, 2, 3)
169 4, 3, 15) 252 ANY11(ARGBToARGB1555Row_Any_SSE2, ARGBToARGB1555Row_SSE2, 0, 4, 2, 3)
170 RGBANY(ARGBToRGB565Row_Any_SSE2, ARGBToRGB565Row_SSE2, ARGBToRGB565Row_C, 253 ANY11(ARGBToARGB4444Row_Any_SSE2, ARGBToARGB4444Row_SSE2, 0, 4, 2, 3)
171 4, 2, 3)
172 RGBANY(ARGBToARGB1555Row_Any_SSE2, ARGBToARGB1555Row_SSE2, ARGBToARGB1555Row_C,
173 4, 2, 3)
174 RGBANY(ARGBToARGB4444Row_Any_SSE2, ARGBToARGB4444Row_SSE2, ARGBToARGB4444Row_C,
175 4, 2, 3)
176 #endif 254 #endif
177 #if defined(HAS_ARGBTOARGB4444ROW_AVX2) 255 #if defined(HAS_ARGBTOARGB4444ROW_AVX2)
178 RGBANY(ARGBToRGB565Row_Any_AVX2, ARGBToRGB565Row_AVX2, ARGBToRGB565Row_C, 256 ANY11(ARGBToRGB565Row_Any_AVX2, ARGBToRGB565Row_AVX2, 0, 4, 2, 7)
179 4, 2, 7) 257 ANY11(ARGBToARGB1555Row_Any_AVX2, ARGBToARGB1555Row_AVX2, 0, 4, 2, 7)
180 RGBANY(ARGBToARGB1555Row_Any_AVX2, ARGBToARGB1555Row_AVX2, ARGBToARGB1555Row_C, 258 ANY11(ARGBToARGB4444Row_Any_AVX2, ARGBToARGB4444Row_AVX2, 0, 4, 2, 7)
181 4, 2, 7) 259 #endif
182 RGBANY(ARGBToARGB4444Row_Any_AVX2, ARGBToARGB4444Row_AVX2, ARGBToARGB4444Row_C, 260 #if defined(HAS_J400TOARGBROW_SSE2)
183 4, 2, 7) 261 ANY11(J400ToARGBRow_Any_SSE2, J400ToARGBRow_SSE2, 0, 1, 4, 7)
184 #endif 262 #endif
185 263 #if defined(HAS_J400TOARGBROW_AVX2)
264 ANY11(J400ToARGBRow_Any_AVX2, J400ToARGBRow_AVX2, 0, 1, 4, 15)
265 #endif
186 #if defined(HAS_I400TOARGBROW_SSE2) 266 #if defined(HAS_I400TOARGBROW_SSE2)
187 RGBANY(I400ToARGBRow_Any_SSE2, I400ToARGBRow_SSE2, I400ToARGBRow_C, 1, 4, 7) 267 ANY11(I400ToARGBRow_Any_SSE2, I400ToARGBRow_SSE2, 0, 1, 4, 7)
188 #endif 268 #endif
189 #if defined(HAS_YTOARGBROW_SSE2) 269 #if defined(HAS_I400TOARGBROW_AVX2)
190 RGBANY(YToARGBRow_Any_SSE2, YToARGBRow_SSE2, YToARGBRow_C, 1, 4, 7) 270 ANY11(I400ToARGBRow_Any_AVX2, I400ToARGBRow_AVX2, 0, 1, 4, 15)
191 #endif
192 #if defined(HAS_YTOARGBROW_AVX2)
193 RGBANY(YToARGBRow_Any_AVX2, YToARGBRow_AVX2, YToARGBRow_C, 1, 4, 15)
194 #endif 271 #endif
195 #if defined(HAS_YUY2TOARGBROW_SSSE3) 272 #if defined(HAS_YUY2TOARGBROW_SSSE3)
196 RGBANY(YUY2ToARGBRow_Any_SSSE3, YUY2ToARGBRow_SSSE3, YUY2ToARGBRow_C, 2, 4, 15) 273 ANY11(YUY2ToARGBRow_Any_SSSE3, YUY2ToARGBRow_SSSE3, 1, 4, 4, 15)
197 RGBANY(UYVYToARGBRow_Any_SSSE3, UYVYToARGBRow_SSSE3, UYVYToARGBRow_C, 2, 4, 15) 274 ANY11(UYVYToARGBRow_Any_SSSE3, UYVYToARGBRow_SSSE3, 1, 4, 4, 15)
198 RGBANY(RGB24ToARGBRow_Any_SSSE3, RGB24ToARGBRow_SSSE3, RGB24ToARGBRow_C, 275 ANY11(RGB24ToARGBRow_Any_SSSE3, RGB24ToARGBRow_SSSE3, 0, 3, 4, 15)
199 3, 4, 15) 276 ANY11(RAWToARGBRow_Any_SSSE3, RAWToARGBRow_SSSE3, 0, 3, 4, 15)
200 RGBANY(RAWToARGBRow_Any_SSSE3, RAWToARGBRow_SSSE3, RAWToARGBRow_C, 3, 4, 15) 277 ANY11(RGB565ToARGBRow_Any_SSE2, RGB565ToARGBRow_SSE2, 0, 2, 4, 7)
201 RGBANY(RGB565ToARGBRow_Any_SSE2, RGB565ToARGBRow_SSE2, RGB565ToARGBRow_C, 278 ANY11(ARGB1555ToARGBRow_Any_SSE2, ARGB1555ToARGBRow_SSE2, 0, 2, 4, 7)
202 2, 4, 7) 279 ANY11(ARGB4444ToARGBRow_Any_SSE2, ARGB4444ToARGBRow_SSE2, 0, 2, 4, 7)
203 RGBANY(ARGB1555ToARGBRow_Any_SSE2, ARGB1555ToARGBRow_SSE2, ARGB1555ToARGBRow_C, 280 #endif
204 2, 4, 7) 281 #if defined(HAS_RGB565TOARGBROW_AVX2)
205 RGBANY(ARGB4444ToARGBRow_Any_SSE2, ARGB4444ToARGBRow_SSE2, ARGB4444ToARGBRow_C, 282 ANY11(RGB565ToARGBRow_Any_AVX2, RGB565ToARGBRow_AVX2, 0, 2, 4, 15)
206 2, 4, 7) 283 #endif
284 #if defined(HAS_ARGB1555TOARGBROW_AVX2)
285 ANY11(ARGB1555ToARGBRow_Any_AVX2, ARGB1555ToARGBRow_AVX2, 0, 2, 4, 15)
286 #endif
287 #if defined(HAS_ARGB4444TOARGBROW_AVX2)
288 ANY11(ARGB4444ToARGBRow_Any_AVX2, ARGB4444ToARGBRow_AVX2, 0, 2, 4, 15)
207 #endif 289 #endif
208 #if defined(HAS_YUY2TOARGBROW_AVX2) 290 #if defined(HAS_YUY2TOARGBROW_AVX2)
209 RGBANY(YUY2ToARGBRow_Any_AVX2, YUY2ToARGBRow_AVX2, YUY2ToARGBRow_C, 2, 4, 31) 291 ANY11(YUY2ToARGBRow_Any_AVX2, YUY2ToARGBRow_AVX2, 1, 4, 4, 31)
210 RGBANY(UYVYToARGBRow_Any_AVX2, UYVYToARGBRow_AVX2, UYVYToARGBRow_C, 2, 4, 31) 292 ANY11(UYVYToARGBRow_Any_AVX2, UYVYToARGBRow_AVX2, 1, 4, 4, 31)
211 #endif 293 #endif
212 #if defined(HAS_ARGBTORGB24ROW_NEON) 294 #if defined(HAS_ARGBTORGB24ROW_NEON)
213 RGBANY(ARGBToRGB24Row_Any_NEON, ARGBToRGB24Row_NEON, ARGBToRGB24Row_C, 4, 3, 7) 295 ANY11(ARGBToRGB24Row_Any_NEON, ARGBToRGB24Row_NEON, 0, 4, 3, 7)
214 RGBANY(ARGBToRAWRow_Any_NEON, ARGBToRAWRow_NEON, ARGBToRAWRow_C, 4, 3, 7) 296 ANY11(ARGBToRAWRow_Any_NEON, ARGBToRAWRow_NEON, 0, 4, 3, 7)
215 RGBANY(ARGBToRGB565Row_Any_NEON, ARGBToRGB565Row_NEON, ARGBToRGB565Row_C, 297 ANY11(ARGBToRGB565Row_Any_NEON, ARGBToRGB565Row_NEON, 0, 4, 2, 7)
216 4, 2, 7) 298 ANY11(ARGBToARGB1555Row_Any_NEON, ARGBToARGB1555Row_NEON, 0, 4, 2, 7)
217 RGBANY(ARGBToARGB1555Row_Any_NEON, ARGBToARGB1555Row_NEON, ARGBToARGB1555Row_C, 299 ANY11(ARGBToARGB4444Row_Any_NEON, ARGBToARGB4444Row_NEON, 0, 4, 2, 7)
218 4, 2, 7) 300 ANY11(J400ToARGBRow_Any_NEON, J400ToARGBRow_NEON, 0, 1, 4, 7)
219 RGBANY(ARGBToARGB4444Row_Any_NEON, ARGBToARGB4444Row_NEON, ARGBToARGB4444Row_C, 301 ANY11(I400ToARGBRow_Any_NEON, I400ToARGBRow_NEON, 0, 1, 4, 7)
220 4, 2, 7) 302 ANY11(YUY2ToARGBRow_Any_NEON, YUY2ToARGBRow_NEON, 1, 4, 4, 7)
221 RGBANY(I400ToARGBRow_Any_NEON, I400ToARGBRow_NEON, I400ToARGBRow_C, 1, 4, 7) 303 ANY11(UYVYToARGBRow_Any_NEON, UYVYToARGBRow_NEON, 1, 4, 4, 7)
222 RGBANY(YToARGBRow_Any_NEON, YToARGBRow_NEON, YToARGBRow_C, 1, 4, 7) 304 #endif
223 RGBANY(YUY2ToARGBRow_Any_NEON, YUY2ToARGBRow_NEON, YUY2ToARGBRow_C, 2, 4, 7)
224 RGBANY(UYVYToARGBRow_Any_NEON, UYVYToARGBRow_NEON, UYVYToARGBRow_C, 2, 4, 7)
225 #endif
226 #undef RGBANY
227
228 // ARGB to Bayer does multiple of 4 pixels, SSSE3 aligned src, unaligned dst.
229 #define BAYERANY(NAMEANY, ARGBTORGB_SIMD, ARGBTORGB_C, SBPP, BPP, MASK) \
230 void NAMEANY(const uint8* src, uint8* dst, uint32 selector, int width) { \
231 int n = width & ~MASK; \
232 if (n > 0) { \
233 ARGBTORGB_SIMD(src, dst, selector, n); \
234 } \
235 ARGBTORGB_C(src + n * SBPP, dst + n * BPP, selector, width & MASK); \
236 }
237
238 #if defined(HAS_ARGBTOBAYERGGROW_SSE2)
239 BAYERANY(ARGBToBayerGGRow_Any_SSE2, ARGBToBayerGGRow_SSE2, ARGBToBayerGGRow_C,
240 4, 1, 7)
241 #endif
242 #if defined(HAS_ARGBTOBAYERGGROW_NEON)
243 BAYERANY(ARGBToBayerGGRow_Any_NEON, ARGBToBayerGGRow_NEON, ARGBToBayerGGRow_C,
244 4, 1, 7)
245 #endif
246
247 #undef BAYERANY
248
249 #define YANY(NAMEANY, ARGBTOY_SIMD, ARGBTOY_C, SBPP, BPP, MASK) \
250 void NAMEANY(const uint8* src_argb, uint8* dst_y, int width) { \
251 int n = width & ~MASK; \
252 if (n > 0) { \
253 ARGBTOY_SIMD(src_argb, dst_y, n); \
254 } \
255 ARGBTOY_C(src_argb + n * SBPP, \
256 dst_y + n * BPP, width & MASK); \
257 }
258 #ifdef HAS_ARGBTOYROW_AVX2 305 #ifdef HAS_ARGBTOYROW_AVX2
259 YANY(ARGBToYRow_Any_AVX2, ARGBToYRow_AVX2, ARGBToYRow_C, 4, 1, 31) 306 ANY11(ARGBToYRow_Any_AVX2, ARGBToYRow_AVX2, 0, 4, 1, 31)
260 #endif 307 #endif
261 #ifdef HAS_ARGBTOYJROW_AVX2 308 #ifdef HAS_ARGBTOYJROW_AVX2
262 YANY(ARGBToYJRow_Any_AVX2, ARGBToYJRow_AVX2, ARGBToYJRow_C, 4, 1, 31) 309 ANY11(ARGBToYJRow_Any_AVX2, ARGBToYJRow_AVX2, 0, 4, 1, 31)
263 #endif 310 #endif
264 #ifdef HAS_UYVYTOYROW_AVX2 311 #ifdef HAS_UYVYTOYROW_AVX2
265 YANY(UYVYToYRow_Any_AVX2, UYVYToYRow_AVX2, UYVYToYRow_C, 2, 1, 31) 312 ANY11(UYVYToYRow_Any_AVX2, UYVYToYRow_AVX2, 0, 2, 1, 31)
266 #endif 313 #endif
267 #ifdef HAS_YUY2TOYROW_AVX2 314 #ifdef HAS_YUY2TOYROW_AVX2
268 YANY(YUY2ToYRow_Any_AVX2, YUY2ToYRow_AVX2, YUY2ToYRow_C, 2, 1, 31) 315 ANY11(YUY2ToYRow_Any_AVX2, YUY2ToYRow_AVX2, 1, 4, 1, 31)
269 #endif 316 #endif
270 #ifdef HAS_ARGBTOYROW_SSSE3 317 #ifdef HAS_ARGBTOYROW_SSSE3
271 YANY(ARGBToYRow_Any_SSSE3, ARGBToYRow_SSSE3, ARGBToYRow_C, 4, 1, 15) 318 ANY11(ARGBToYRow_Any_SSSE3, ARGBToYRow_SSSE3, 0, 4, 1, 15)
272 #endif 319 #endif
273 #ifdef HAS_BGRATOYROW_SSSE3 320 #ifdef HAS_BGRATOYROW_SSSE3
274 YANY(BGRAToYRow_Any_SSSE3, BGRAToYRow_SSSE3, BGRAToYRow_C, 4, 1, 15) 321 ANY11(BGRAToYRow_Any_SSSE3, BGRAToYRow_SSSE3, 0, 4, 1, 15)
275 YANY(ABGRToYRow_Any_SSSE3, ABGRToYRow_SSSE3, ABGRToYRow_C, 4, 1, 15) 322 ANY11(ABGRToYRow_Any_SSSE3, ABGRToYRow_SSSE3, 0, 4, 1, 15)
276 YANY(RGBAToYRow_Any_SSSE3, RGBAToYRow_SSSE3, RGBAToYRow_C, 4, 1, 15) 323 ANY11(RGBAToYRow_Any_SSSE3, RGBAToYRow_SSSE3, 0, 4, 1, 15)
277 YANY(YUY2ToYRow_Any_SSE2, YUY2ToYRow_SSE2, YUY2ToYRow_C, 2, 1, 15) 324 ANY11(YUY2ToYRow_Any_SSE2, YUY2ToYRow_SSE2, 1, 4, 1, 15)
278 YANY(UYVYToYRow_Any_SSE2, UYVYToYRow_SSE2, UYVYToYRow_C, 2, 1, 15) 325 ANY11(UYVYToYRow_Any_SSE2, UYVYToYRow_SSE2, 1, 4, 1, 15)
279 #endif 326 #endif
280 #ifdef HAS_ARGBTOYJROW_SSSE3 327 #ifdef HAS_ARGBTOYJROW_SSSE3
281 YANY(ARGBToYJRow_Any_SSSE3, ARGBToYJRow_SSSE3, ARGBToYJRow_C, 4, 1, 15) 328 ANY11(ARGBToYJRow_Any_SSSE3, ARGBToYJRow_SSSE3, 0, 4, 1, 15)
282 #endif 329 #endif
283 #ifdef HAS_ARGBTOYROW_NEON 330 #ifdef HAS_ARGBTOYROW_NEON
284 YANY(ARGBToYRow_Any_NEON, ARGBToYRow_NEON, ARGBToYRow_C, 4, 1, 7) 331 ANY11(ARGBToYRow_Any_NEON, ARGBToYRow_NEON, 0, 4, 1, 7)
285 #endif 332 #endif
286 #ifdef HAS_ARGBTOYJROW_NEON 333 #ifdef HAS_ARGBTOYJROW_NEON
287 YANY(ARGBToYJRow_Any_NEON, ARGBToYJRow_NEON, ARGBToYJRow_C, 4, 1, 7) 334 ANY11(ARGBToYJRow_Any_NEON, ARGBToYJRow_NEON, 0, 4, 1, 7)
288 #endif 335 #endif
289 #ifdef HAS_BGRATOYROW_NEON 336 #ifdef HAS_BGRATOYROW_NEON
290 YANY(BGRAToYRow_Any_NEON, BGRAToYRow_NEON, BGRAToYRow_C, 4, 1, 7) 337 ANY11(BGRAToYRow_Any_NEON, BGRAToYRow_NEON, 0, 4, 1, 7)
291 #endif 338 #endif
292 #ifdef HAS_ABGRTOYROW_NEON 339 #ifdef HAS_ABGRTOYROW_NEON
293 YANY(ABGRToYRow_Any_NEON, ABGRToYRow_NEON, ABGRToYRow_C, 4, 1, 7) 340 ANY11(ABGRToYRow_Any_NEON, ABGRToYRow_NEON, 0, 4, 1, 7)
294 #endif 341 #endif
295 #ifdef HAS_RGBATOYROW_NEON 342 #ifdef HAS_RGBATOYROW_NEON
296 YANY(RGBAToYRow_Any_NEON, RGBAToYRow_NEON, RGBAToYRow_C, 4, 1, 7) 343 ANY11(RGBAToYRow_Any_NEON, RGBAToYRow_NEON, 0, 4, 1, 7)
297 #endif 344 #endif
298 #ifdef HAS_RGB24TOYROW_NEON 345 #ifdef HAS_RGB24TOYROW_NEON
299 YANY(RGB24ToYRow_Any_NEON, RGB24ToYRow_NEON, RGB24ToYRow_C, 3, 1, 7) 346 ANY11(RGB24ToYRow_Any_NEON, RGB24ToYRow_NEON, 0, 3, 1, 7)
300 #endif 347 #endif
301 #ifdef HAS_RAWTOYROW_NEON 348 #ifdef HAS_RAWTOYROW_NEON
302 YANY(RAWToYRow_Any_NEON, RAWToYRow_NEON, RAWToYRow_C, 3, 1, 7) 349 ANY11(RAWToYRow_Any_NEON, RAWToYRow_NEON, 0, 3, 1, 7)
303 #endif 350 #endif
304 #ifdef HAS_RGB565TOYROW_NEON 351 #ifdef HAS_RGB565TOYROW_NEON
305 YANY(RGB565ToYRow_Any_NEON, RGB565ToYRow_NEON, RGB565ToYRow_C, 2, 1, 7) 352 ANY11(RGB565ToYRow_Any_NEON, RGB565ToYRow_NEON, 0, 2, 1, 7)
306 #endif 353 #endif
307 #ifdef HAS_ARGB1555TOYROW_NEON 354 #ifdef HAS_ARGB1555TOYROW_NEON
308 YANY(ARGB1555ToYRow_Any_NEON, ARGB1555ToYRow_NEON, ARGB1555ToYRow_C, 2, 1, 7) 355 ANY11(ARGB1555ToYRow_Any_NEON, ARGB1555ToYRow_NEON, 0, 2, 1, 7)
309 #endif 356 #endif
310 #ifdef HAS_ARGB4444TOYROW_NEON 357 #ifdef HAS_ARGB4444TOYROW_NEON
311 YANY(ARGB4444ToYRow_Any_NEON, ARGB4444ToYRow_NEON, ARGB4444ToYRow_C, 2, 1, 7) 358 ANY11(ARGB4444ToYRow_Any_NEON, ARGB4444ToYRow_NEON, 0, 2, 1, 7)
312 #endif 359 #endif
313 #ifdef HAS_YUY2TOYROW_NEON 360 #ifdef HAS_YUY2TOYROW_NEON
314 YANY(YUY2ToYRow_Any_NEON, YUY2ToYRow_NEON, YUY2ToYRow_C, 2, 1, 15) 361 ANY11(YUY2ToYRow_Any_NEON, YUY2ToYRow_NEON, 1, 4, 1, 15)
315 #endif 362 #endif
316 #ifdef HAS_UYVYTOYROW_NEON 363 #ifdef HAS_UYVYTOYROW_NEON
317 YANY(UYVYToYRow_Any_NEON, UYVYToYRow_NEON, UYVYToYRow_C, 2, 1, 15) 364 ANY11(UYVYToYRow_Any_NEON, UYVYToYRow_NEON, 0, 2, 1, 15)
318 #endif 365 #endif
319 #ifdef HAS_RGB24TOARGBROW_NEON 366 #ifdef HAS_RGB24TOARGBROW_NEON
320 YANY(RGB24ToARGBRow_Any_NEON, RGB24ToARGBRow_NEON, RGB24ToARGBRow_C, 3, 4, 7) 367 ANY11(RGB24ToARGBRow_Any_NEON, RGB24ToARGBRow_NEON, 0, 3, 4, 7)
321 #endif 368 #endif
322 #ifdef HAS_RAWTOARGBROW_NEON 369 #ifdef HAS_RAWTOARGBROW_NEON
323 YANY(RAWToARGBRow_Any_NEON, RAWToARGBRow_NEON, RAWToARGBRow_C, 3, 4, 7) 370 ANY11(RAWToARGBRow_Any_NEON, RAWToARGBRow_NEON, 0, 3, 4, 7)
324 #endif 371 #endif
325 #ifdef HAS_RGB565TOARGBROW_NEON 372 #ifdef HAS_RGB565TOARGBROW_NEON
326 YANY(RGB565ToARGBRow_Any_NEON, RGB565ToARGBRow_NEON, RGB565ToARGBRow_C, 2, 4, 7) 373 ANY11(RGB565ToARGBRow_Any_NEON, RGB565ToARGBRow_NEON, 0, 2, 4, 7)
327 #endif 374 #endif
328 #ifdef HAS_ARGB1555TOARGBROW_NEON 375 #ifdef HAS_ARGB1555TOARGBROW_NEON
329 YANY(ARGB1555ToARGBRow_Any_NEON, ARGB1555ToARGBRow_NEON, ARGB1555ToARGBRow_C, 376 ANY11(ARGB1555ToARGBRow_Any_NEON, ARGB1555ToARGBRow_NEON, 0, 2, 4, 7)
330 2, 4, 7)
331 #endif 377 #endif
332 #ifdef HAS_ARGB4444TOARGBROW_NEON 378 #ifdef HAS_ARGB4444TOARGBROW_NEON
333 YANY(ARGB4444ToARGBRow_Any_NEON, ARGB4444ToARGBRow_NEON, ARGB4444ToARGBRow_C, 379 ANY11(ARGB4444ToARGBRow_Any_NEON, ARGB4444ToARGBRow_NEON, 0, 2, 4, 7)
334 2, 4, 7)
335 #endif 380 #endif
336 #ifdef HAS_ARGBATTENUATEROW_SSSE3 381 #ifdef HAS_ARGBATTENUATEROW_SSSE3
337 YANY(ARGBAttenuateRow_Any_SSSE3, ARGBAttenuateRow_SSSE3, ARGBAttenuateRow_C, 382 ANY11(ARGBAttenuateRow_Any_SSSE3, ARGBAttenuateRow_SSSE3, 0, 4, 4, 3)
338 4, 4, 3)
339 #endif 383 #endif
340 #ifdef HAS_ARGBATTENUATEROW_SSE2 384 #ifdef HAS_ARGBATTENUATEROW_SSE2
341 YANY(ARGBAttenuateRow_Any_SSE2, ARGBAttenuateRow_SSE2, ARGBAttenuateRow_C, 385 ANY11(ARGBAttenuateRow_Any_SSE2, ARGBAttenuateRow_SSE2, 0, 4, 4, 3)
342 4, 4, 3)
343 #endif 386 #endif
344 #ifdef HAS_ARGBUNATTENUATEROW_SSE2 387 #ifdef HAS_ARGBUNATTENUATEROW_SSE2
345 YANY(ARGBUnattenuateRow_Any_SSE2, ARGBUnattenuateRow_SSE2, ARGBUnattenuateRow_C, 388 ANY11(ARGBUnattenuateRow_Any_SSE2, ARGBUnattenuateRow_SSE2, 0, 4, 4, 3)
346 4, 4, 3)
347 #endif 389 #endif
348 #ifdef HAS_ARGBATTENUATEROW_AVX2 390 #ifdef HAS_ARGBATTENUATEROW_AVX2
349 YANY(ARGBAttenuateRow_Any_AVX2, ARGBAttenuateRow_AVX2, ARGBAttenuateRow_C, 391 ANY11(ARGBAttenuateRow_Any_AVX2, ARGBAttenuateRow_AVX2, 0, 4, 4, 7)
350 4, 4, 7)
351 #endif 392 #endif
352 #ifdef HAS_ARGBUNATTENUATEROW_AVX2 393 #ifdef HAS_ARGBUNATTENUATEROW_AVX2
353 YANY(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, ARGBUnattenuateRow_C, 394 ANY11(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, 0, 4, 4, 7)
354 4, 4, 7)
355 #endif 395 #endif
356 #ifdef HAS_ARGBATTENUATEROW_NEON 396 #ifdef HAS_ARGBATTENUATEROW_NEON
357 YANY(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, ARGBAttenuateRow_C, 397 ANY11(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, 0, 4, 4, 7)
358 4, 4, 7) 398 #endif
359 #endif 399 #undef ANY11
360 #undef YANY 400
361 401 // Any 1 to 1 with parameter.
362 // RGB/YUV to UV does multiple of 16 with SIMD and remainder with C. 402 #define ANY11P(NAMEANY, ANY_SIMD, T, SBPP, BPP, MASK) \
363 #define UVANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, MASK) \ 403 void NAMEANY(const uint8* src_ptr, uint8* dst_ptr, \
364 void NAMEANY(const uint8* src_argb, int src_stride_argb, \ 404 T shuffler, int width) { \
365 uint8* dst_u, uint8* dst_v, int width) { \ 405 SIMD_ALIGNED(uint8 temp[64 * 2]); \
366 int n = width & ~MASK; \ 406 memset(temp, 0, 64); /* for msan */ \
367 if (n > 0) { \ 407 int r = width & MASK; \
368 ANYTOUV_SIMD(src_argb, src_stride_argb, dst_u, dst_v, n); \ 408 int n = width & ~MASK; \
369 } \ 409 if (n > 0) { \
370 ANYTOUV_C(src_argb + n * BPP, src_stride_argb, \ 410 ANY_SIMD(src_ptr, dst_ptr, shuffler, n); \
371 dst_u + (n >> 1), \ 411 } \
372 dst_v + (n >> 1), \ 412 memcpy(temp, src_ptr + n * SBPP, r * SBPP); \
373 width & MASK); \ 413 ANY_SIMD(temp, temp + 64, shuffler, MASK + 1); \
374 } 414 memcpy(dst_ptr + n * BPP, temp + 64, r * BPP); \
375 415 }
376 #ifdef HAS_ARGBTOUVROW_AVX2 416
377 UVANY(ARGBToUVRow_Any_AVX2, ARGBToUVRow_AVX2, ARGBToUVRow_C, 4, 31) 417 #if defined(HAS_ARGBTORGB565DITHERROW_SSE2)
378 #endif 418 ANY11P(ARGBToRGB565DitherRow_Any_SSE2, ARGBToRGB565DitherRow_SSE2,
379 #ifdef HAS_ARGBTOUVROW_SSSE3 419 const uint32, 4, 2, 3)
380 UVANY(ARGBToUVRow_Any_SSSE3, ARGBToUVRow_SSSE3, ARGBToUVRow_C, 4, 15) 420 #endif
381 UVANY(ARGBToUVJRow_Any_SSSE3, ARGBToUVJRow_SSSE3, ARGBToUVJRow_C, 4, 15) 421 #if defined(HAS_ARGBTORGB565DITHERROW_AVX2)
382 UVANY(BGRAToUVRow_Any_SSSE3, BGRAToUVRow_SSSE3, BGRAToUVRow_C, 4, 15) 422 ANY11P(ARGBToRGB565DitherRow_Any_AVX2, ARGBToRGB565DitherRow_AVX2,
383 UVANY(ABGRToUVRow_Any_SSSE3, ABGRToUVRow_SSSE3, ABGRToUVRow_C, 4, 15) 423 const uint32, 4, 2, 7)
384 UVANY(RGBAToUVRow_Any_SSSE3, RGBAToUVRow_SSSE3, RGBAToUVRow_C, 4, 15) 424 #endif
385 #endif 425 #if defined(HAS_ARGBTORGB565DITHERROW_NEON)
386 #ifdef HAS_YUY2TOUVROW_AVX2 426 ANY11P(ARGBToRGB565DitherRow_Any_NEON, ARGBToRGB565DitherRow_NEON,
387 UVANY(YUY2ToUVRow_Any_AVX2, YUY2ToUVRow_AVX2, YUY2ToUVRow_C, 2, 31) 427 const uint32, 4, 2, 7)
388 UVANY(UYVYToUVRow_Any_AVX2, UYVYToUVRow_AVX2, UYVYToUVRow_C, 2, 31) 428 #endif
389 #endif
390 #ifdef HAS_YUY2TOUVROW_SSE2
391 UVANY(YUY2ToUVRow_Any_SSE2, YUY2ToUVRow_SSE2, YUY2ToUVRow_C, 2, 15)
392 UVANY(UYVYToUVRow_Any_SSE2, UYVYToUVRow_SSE2, UYVYToUVRow_C, 2, 15)
393 #endif
394 #ifdef HAS_ARGBTOUVROW_NEON
395 UVANY(ARGBToUVRow_Any_NEON, ARGBToUVRow_NEON, ARGBToUVRow_C, 4, 15)
396 #endif
397 #ifdef HAS_ARGBTOUVJROW_NEON
398 UVANY(ARGBToUVJRow_Any_NEON, ARGBToUVJRow_NEON, ARGBToUVJRow_C, 4, 15)
399 #endif
400 #ifdef HAS_BGRATOUVROW_NEON
401 UVANY(BGRAToUVRow_Any_NEON, BGRAToUVRow_NEON, BGRAToUVRow_C, 4, 15)
402 #endif
403 #ifdef HAS_ABGRTOUVROW_NEON
404 UVANY(ABGRToUVRow_Any_NEON, ABGRToUVRow_NEON, ABGRToUVRow_C, 4, 15)
405 #endif
406 #ifdef HAS_RGBATOUVROW_NEON
407 UVANY(RGBAToUVRow_Any_NEON, RGBAToUVRow_NEON, RGBAToUVRow_C, 4, 15)
408 #endif
409 #ifdef HAS_RGB24TOUVROW_NEON
410 UVANY(RGB24ToUVRow_Any_NEON, RGB24ToUVRow_NEON, RGB24ToUVRow_C, 3, 15)
411 #endif
412 #ifdef HAS_RAWTOUVROW_NEON
413 UVANY(RAWToUVRow_Any_NEON, RAWToUVRow_NEON, RAWToUVRow_C, 3, 15)
414 #endif
415 #ifdef HAS_RGB565TOUVROW_NEON
416 UVANY(RGB565ToUVRow_Any_NEON, RGB565ToUVRow_NEON, RGB565ToUVRow_C, 2, 15)
417 #endif
418 #ifdef HAS_ARGB1555TOUVROW_NEON
419 UVANY(ARGB1555ToUVRow_Any_NEON, ARGB1555ToUVRow_NEON, ARGB1555ToUVRow_C, 2, 15)
420 #endif
421 #ifdef HAS_ARGB4444TOUVROW_NEON
422 UVANY(ARGB4444ToUVRow_Any_NEON, ARGB4444ToUVRow_NEON, ARGB4444ToUVRow_C, 2, 15)
423 #endif
424 #ifdef HAS_YUY2TOUVROW_NEON
425 UVANY(YUY2ToUVRow_Any_NEON, YUY2ToUVRow_NEON, YUY2ToUVRow_C, 2, 15)
426 #endif
427 #ifdef HAS_UYVYTOUVROW_NEON
428 UVANY(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, UYVYToUVRow_C, 2, 15)
429 #endif
430 #undef UVANY
431
432 #define UV422ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, BPP, SHIFT, MASK) \
433 void NAMEANY(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int width) { \
434 int n = width & ~MASK; \
435 if (n > 0) { \
436 ANYTOUV_SIMD(src_uv, dst_u, dst_v, n); \
437 } \
438 ANYTOUV_C(src_uv + n * BPP, \
439 dst_u + (n >> SHIFT), \
440 dst_v + (n >> SHIFT), \
441 width & MASK); \
442 }
443
444 #ifdef HAS_ARGBTOUV444ROW_SSSE3
445 UV422ANY(ARGBToUV444Row_Any_SSSE3, ARGBToUV444Row_SSSE3,
446 ARGBToUV444Row_C, 4, 0, 15)
447 #endif
448 #ifdef HAS_YUY2TOUV422ROW_AVX2
449 UV422ANY(YUY2ToUV422Row_Any_AVX2, YUY2ToUV422Row_AVX2,
450 YUY2ToUV422Row_C, 2, 1, 31)
451 UV422ANY(UYVYToUV422Row_Any_AVX2, UYVYToUV422Row_AVX2,
452 UYVYToUV422Row_C, 2, 1, 31)
453 #endif
454 #ifdef HAS_ARGBTOUV422ROW_SSSE3
455 UV422ANY(ARGBToUV422Row_Any_SSSE3, ARGBToUV422Row_SSSE3,
456 ARGBToUV422Row_C, 4, 1, 15)
457 #endif
458 #ifdef HAS_YUY2TOUV422ROW_SSE2
459 UV422ANY(YUY2ToUV422Row_Any_SSE2, YUY2ToUV422Row_SSE2,
460 YUY2ToUV422Row_C, 2, 1, 15)
461 UV422ANY(UYVYToUV422Row_Any_SSE2, UYVYToUV422Row_SSE2,
462 UYVYToUV422Row_C, 2, 1, 15)
463 #endif
464 #ifdef HAS_YUY2TOUV422ROW_NEON
465 UV422ANY(ARGBToUV444Row_Any_NEON, ARGBToUV444Row_NEON,
466 ARGBToUV444Row_C, 4, 0, 7)
467 UV422ANY(ARGBToUV422Row_Any_NEON, ARGBToUV422Row_NEON,
468 ARGBToUV422Row_C, 4, 1, 15)
469 UV422ANY(ARGBToUV411Row_Any_NEON, ARGBToUV411Row_NEON,
470 ARGBToUV411Row_C, 4, 2, 31)
471 UV422ANY(YUY2ToUV422Row_Any_NEON, YUY2ToUV422Row_NEON,
472 YUY2ToUV422Row_C, 2, 1, 15)
473 UV422ANY(UYVYToUV422Row_Any_NEON, UYVYToUV422Row_NEON,
474 UYVYToUV422Row_C, 2, 1, 15)
475 #endif
476 #undef UV422ANY
477
478 #define SPLITUVROWANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, MASK) \
479 void NAMEANY(const uint8* src_uv, uint8* dst_u, uint8* dst_v, int width) { \
480 int n = width & ~MASK; \
481 if (n > 0) { \
482 ANYTOUV_SIMD(src_uv, dst_u, dst_v, n); \
483 } \
484 ANYTOUV_C(src_uv + n * 2, \
485 dst_u + n, \
486 dst_v + n, \
487 width & MASK); \
488 }
489
490 #ifdef HAS_SPLITUVROW_SSE2
491 SPLITUVROWANY(SplitUVRow_Any_SSE2, SplitUVRow_SSE2, SplitUVRow_C, 15)
492 #endif
493 #ifdef HAS_SPLITUVROW_AVX2
494 SPLITUVROWANY(SplitUVRow_Any_AVX2, SplitUVRow_AVX2, SplitUVRow_C, 31)
495 #endif
496 #ifdef HAS_SPLITUVROW_NEON
497 SPLITUVROWANY(SplitUVRow_Any_NEON, SplitUVRow_NEON, SplitUVRow_C, 15)
498 #endif
499 #ifdef HAS_SPLITUVROW_MIPS_DSPR2
500 SPLITUVROWANY(SplitUVRow_Any_MIPS_DSPR2, SplitUVRow_MIPS_DSPR2,
501 SplitUVRow_C, 15)
502 #endif
503 #undef SPLITUVROWANY
504
505 #define MERGEUVROW_ANY(NAMEANY, ANYTOUV_SIMD, ANYTOUV_C, MASK) \
506 void NAMEANY(const uint8* src_u, const uint8* src_v, \
507 uint8* dst_uv, int width) { \
508 int n = width & ~MASK; \
509 if (n > 0) { \
510 ANYTOUV_SIMD(src_u, src_v, dst_uv, n); \
511 } \
512 ANYTOUV_C(src_u + n, \
513 src_v + n, \
514 dst_uv + n * 2, \
515 width & MASK); \
516 }
517
518 #ifdef HAS_MERGEUVROW_SSE2
519 MERGEUVROW_ANY(MergeUVRow_Any_SSE2, MergeUVRow_SSE2, MergeUVRow_C, 15)
520 #endif
521 #ifdef HAS_MERGEUVROW_AVX2
522 MERGEUVROW_ANY(MergeUVRow_Any_AVX2, MergeUVRow_AVX2, MergeUVRow_C, 31)
523 #endif
524 #ifdef HAS_MERGEUVROW_NEON
525 MERGEUVROW_ANY(MergeUVRow_Any_NEON, MergeUVRow_NEON, MergeUVRow_C, 15)
526 #endif
527 #undef MERGEUVROW_ANY
528
529 #define MATHROW_ANY(NAMEANY, ARGBMATH_SIMD, ARGBMATH_C, MASK) \
530 void NAMEANY(const uint8* src_argb0, const uint8* src_argb1, \
531 uint8* dst_argb, int width) { \
532 int n = width & ~MASK; \
533 if (n > 0) { \
534 ARGBMATH_SIMD(src_argb0, src_argb1, dst_argb, n); \
535 } \
536 ARGBMATH_C(src_argb0 + n * 4, \
537 src_argb1 + n * 4, \
538 dst_argb + n * 4, \
539 width & MASK); \
540 }
541
542 #ifdef HAS_ARGBMULTIPLYROW_SSE2
543 MATHROW_ANY(ARGBMultiplyRow_Any_SSE2, ARGBMultiplyRow_SSE2, ARGBMultiplyRow_C,
544 3)
545 #endif
546 #ifdef HAS_ARGBADDROW_SSE2
547 MATHROW_ANY(ARGBAddRow_Any_SSE2, ARGBAddRow_SSE2, ARGBAddRow_C, 3)
548 #endif
549 #ifdef HAS_ARGBSUBTRACTROW_SSE2
550 MATHROW_ANY(ARGBSubtractRow_Any_SSE2, ARGBSubtractRow_SSE2, ARGBSubtractRow_C,
551 3)
552 #endif
553 #ifdef HAS_ARGBMULTIPLYROW_AVX2
554 MATHROW_ANY(ARGBMultiplyRow_Any_AVX2, ARGBMultiplyRow_AVX2, ARGBMultiplyRow_C,
555 7)
556 #endif
557 #ifdef HAS_ARGBADDROW_AVX2
558 MATHROW_ANY(ARGBAddRow_Any_AVX2, ARGBAddRow_AVX2, ARGBAddRow_C, 7)
559 #endif
560 #ifdef HAS_ARGBSUBTRACTROW_AVX2
561 MATHROW_ANY(ARGBSubtractRow_Any_AVX2, ARGBSubtractRow_AVX2, ARGBSubtractRow_C,
562 7)
563 #endif
564 #ifdef HAS_ARGBMULTIPLYROW_NEON
565 MATHROW_ANY(ARGBMultiplyRow_Any_NEON, ARGBMultiplyRow_NEON, ARGBMultiplyRow_C,
566 7)
567 #endif
568 #ifdef HAS_ARGBADDROW_NEON
569 MATHROW_ANY(ARGBAddRow_Any_NEON, ARGBAddRow_NEON, ARGBAddRow_C, 7)
570 #endif
571 #ifdef HAS_ARGBSUBTRACTROW_NEON
572 MATHROW_ANY(ARGBSubtractRow_Any_NEON, ARGBSubtractRow_NEON, ARGBSubtractRow_C,
573 7)
574 #endif
575 #undef MATHROW_ANY
576
577 // Shuffle may want to work in place, so last16 method can not be used.
578 #define YANY(NAMEANY, ARGBTOY_SIMD, ARGBTOY_C, SBPP, BPP, MASK) \
579 void NAMEANY(const uint8* src_argb, uint8* dst_argb, \
580 const uint8* shuffler, int width) { \
581 int n = width & ~MASK; \
582 if (n > 0) { \
583 ARGBTOY_SIMD(src_argb, dst_argb, shuffler, n); \
584 } \
585 ARGBTOY_C(src_argb + n * SBPP, \
586 dst_argb + n * BPP, shuffler, width & MASK); \
587 }
588
589 #ifdef HAS_ARGBSHUFFLEROW_SSE2 429 #ifdef HAS_ARGBSHUFFLEROW_SSE2
590 YANY(ARGBShuffleRow_Any_SSE2, ARGBShuffleRow_SSE2, 430 ANY11P(ARGBShuffleRow_Any_SSE2, ARGBShuffleRow_SSE2, const uint8*, 4, 4, 3)
591 ARGBShuffleRow_C, 4, 4, 3)
592 #endif 431 #endif
593 #ifdef HAS_ARGBSHUFFLEROW_SSSE3 432 #ifdef HAS_ARGBSHUFFLEROW_SSSE3
594 YANY(ARGBShuffleRow_Any_SSSE3, ARGBShuffleRow_SSSE3, 433 ANY11P(ARGBShuffleRow_Any_SSSE3, ARGBShuffleRow_SSSE3, const uint8*, 4, 4, 7)
595 ARGBShuffleRow_C, 4, 4, 7)
596 #endif 434 #endif
597 #ifdef HAS_ARGBSHUFFLEROW_AVX2 435 #ifdef HAS_ARGBSHUFFLEROW_AVX2
598 YANY(ARGBShuffleRow_Any_AVX2, ARGBShuffleRow_AVX2, 436 ANY11P(ARGBShuffleRow_Any_AVX2, ARGBShuffleRow_AVX2, const uint8*, 4, 4, 15)
599 ARGBShuffleRow_C, 4, 4, 15)
600 #endif 437 #endif
601 #ifdef HAS_ARGBSHUFFLEROW_NEON 438 #ifdef HAS_ARGBSHUFFLEROW_NEON
602 YANY(ARGBShuffleRow_Any_NEON, ARGBShuffleRow_NEON, 439 ANY11P(ARGBShuffleRow_Any_NEON, ARGBShuffleRow_NEON, const uint8*, 4, 4, 3)
603 ARGBShuffleRow_C, 4, 4, 3) 440 #endif
604 #endif 441 #undef ANY11P
605 #undef YANY 442
606 443 // Any 1 to 1 interpolate. Takes 2 rows of source via stride.
607 // Interpolate may want to work in place, so last16 method can not be used. 444 #define ANY11T(NAMEANY, ANY_SIMD, SBPP, BPP, MASK) \
608 #define NANY(NAMEANY, TERP_SIMD, TERP_C, SBPP, BPP, MASK) \
609 void NAMEANY(uint8* dst_ptr, const uint8* src_ptr, \ 445 void NAMEANY(uint8* dst_ptr, const uint8* src_ptr, \
610 ptrdiff_t src_stride_ptr, int width, \ 446 ptrdiff_t src_stride_ptr, int width, \
611 int source_y_fraction) { \ 447 int source_y_fraction) { \
612 int n = width & ~MASK; \ 448 SIMD_ALIGNED(uint8 temp[64 * 3]); \
613 if (n > 0) { \ 449 memset(temp, 0, 64 * 2); /* for msan */ \
614 TERP_SIMD(dst_ptr, src_ptr, src_stride_ptr, n, source_y_fraction); \ 450 int r = width & MASK; \
615 } \ 451 int n = width & ~MASK; \
616 TERP_C(dst_ptr + n * BPP, \ 452 if (n > 0) { \
617 src_ptr + n * SBPP, src_stride_ptr, \ 453 ANY_SIMD(dst_ptr, src_ptr, src_stride_ptr, n, source_y_fraction); \
618 width & MASK, source_y_fraction); \ 454 } \
455 memcpy(temp, src_ptr + n * SBPP, r * SBPP); \
456 memcpy(temp + 64, src_ptr + src_stride_ptr + n * SBPP, r * SBPP); \
457 ANY_SIMD(temp + 128, temp, 64, MASK + 1, source_y_fraction); \
458 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \
619 } 459 }
620 460
621 #ifdef HAS_INTERPOLATEROW_AVX2 461 #ifdef HAS_INTERPOLATEROW_AVX2
622 NANY(InterpolateRow_Any_AVX2, InterpolateRow_AVX2, InterpolateRow_C, 1, 1, 31) 462 ANY11T(InterpolateRow_Any_AVX2, InterpolateRow_AVX2, 1, 1, 31)
623 #endif 463 #endif
624 #ifdef HAS_INTERPOLATEROW_SSSE3 464 #ifdef HAS_INTERPOLATEROW_SSSE3
625 NANY(InterpolateRow_Any_SSSE3, InterpolateRow_SSSE3, InterpolateRow_C, 1, 1, 15) 465 ANY11T(InterpolateRow_Any_SSSE3, InterpolateRow_SSSE3, 1, 1, 15)
626 #endif 466 #endif
627 #ifdef HAS_INTERPOLATEROW_SSE2 467 #ifdef HAS_INTERPOLATEROW_SSE2
628 NANY(InterpolateRow_Any_SSE2, InterpolateRow_SSE2, InterpolateRow_C, 1, 1, 15) 468 ANY11T(InterpolateRow_Any_SSE2, InterpolateRow_SSE2, 1, 1, 15)
629 #endif 469 #endif
630 #ifdef HAS_INTERPOLATEROW_NEON 470 #ifdef HAS_INTERPOLATEROW_NEON
631 NANY(InterpolateRow_Any_NEON, InterpolateRow_NEON, InterpolateRow_C, 1, 1, 15) 471 ANY11T(InterpolateRow_Any_NEON, InterpolateRow_NEON, 1, 1, 15)
632 #endif 472 #endif
633 #ifdef HAS_INTERPOLATEROW_MIPS_DSPR2 473 #ifdef HAS_INTERPOLATEROW_MIPS_DSPR2
634 NANY(InterpolateRow_Any_MIPS_DSPR2, InterpolateRow_MIPS_DSPR2, InterpolateRow_C, 474 ANY11T(InterpolateRow_Any_MIPS_DSPR2, InterpolateRow_MIPS_DSPR2, 1, 1, 3)
635 1, 1, 3) 475 #endif
636 #endif 476 #undef ANY11T
637 #undef NANY 477
638 478 // Any 1 to 1 mirror.
639 #define MANY(NAMEANY, MIRROR_SIMD, MIRROR_C, BPP, MASK) \ 479 #define ANY11M(NAMEANY, ANY_SIMD, BPP, MASK) \
640 void NAMEANY(const uint8* src_y, uint8* dst_y, int width) { \ 480 void NAMEANY(const uint8* src_ptr, uint8* dst_ptr, int width) { \
641 int n = width & ~MASK; \ 481 SIMD_ALIGNED(uint8 temp[64 * 2]); \
642 int r = width & MASK; \ 482 memset(temp, 0, 64); /* for msan */ \
643 if (n > 0) { \ 483 int r = width & MASK; \
644 MIRROR_SIMD(src_y, dst_y + r * BPP, n); \ 484 int n = width & ~MASK; \
645 } \ 485 if (n > 0) { \
646 MIRROR_C(src_y + n * BPP, dst_y, r); \ 486 ANY_SIMD(src_ptr + r * BPP, dst_ptr, n); \
487 } \
488 memcpy(temp, src_ptr, r * BPP); \
489 ANY_SIMD(temp, temp + 64, MASK + 1); \
490 memcpy(dst_ptr + n * BPP, temp + 64 + (MASK + 1 - r) * BPP, r * BPP); \
647 } 491 }
648 492
649 #ifdef HAS_MIRRORROW_AVX2 493 #ifdef HAS_MIRRORROW_AVX2
650 MANY(MirrorRow_Any_AVX2, MirrorRow_AVX2, MirrorRow_C, 1, 31) 494 ANY11M(MirrorRow_Any_AVX2, MirrorRow_AVX2, 1, 31)
651 #endif 495 #endif
652 #ifdef HAS_MIRRORROW_SSSE3 496 #ifdef HAS_MIRRORROW_SSSE3
653 MANY(MirrorRow_Any_SSSE3, MirrorRow_SSSE3, MirrorRow_C, 1, 15) 497 ANY11M(MirrorRow_Any_SSSE3, MirrorRow_SSSE3, 1, 15)
654 #endif 498 #endif
655 #ifdef HAS_MIRRORROW_SSE2 499 #ifdef HAS_MIRRORROW_SSE2
656 MANY(MirrorRow_Any_SSE2, MirrorRow_SSE2, MirrorRow_C, 1, 15) 500 ANY11M(MirrorRow_Any_SSE2, MirrorRow_SSE2, 1, 15)
657 #endif 501 #endif
658 #ifdef HAS_MIRRORROW_NEON 502 #ifdef HAS_MIRRORROW_NEON
659 MANY(MirrorRow_Any_NEON, MirrorRow_NEON, MirrorRow_C, 1, 15) 503 ANY11M(MirrorRow_Any_NEON, MirrorRow_NEON, 1, 15)
660 #endif 504 #endif
661 #ifdef HAS_ARGBMIRRORROW_AVX2 505 #ifdef HAS_ARGBMIRRORROW_AVX2
662 MANY(ARGBMirrorRow_Any_AVX2, ARGBMirrorRow_AVX2, ARGBMirrorRow_C, 4, 7) 506 ANY11M(ARGBMirrorRow_Any_AVX2, ARGBMirrorRow_AVX2, 4, 7)
663 #endif 507 #endif
664 #ifdef HAS_ARGBMIRRORROW_SSE2 508 #ifdef HAS_ARGBMIRRORROW_SSE2
665 MANY(ARGBMirrorRow_Any_SSE2, ARGBMirrorRow_SSE2, ARGBMirrorRow_C, 4, 3) 509 ANY11M(ARGBMirrorRow_Any_SSE2, ARGBMirrorRow_SSE2, 4, 3)
666 #endif 510 #endif
667 #ifdef HAS_ARGBMIRRORROW_NEON 511 #ifdef HAS_ARGBMIRRORROW_NEON
668 MANY(ARGBMirrorRow_Any_NEON, ARGBMirrorRow_NEON, ARGBMirrorRow_C, 4, 3) 512 ANY11M(ARGBMirrorRow_Any_NEON, ARGBMirrorRow_NEON, 4, 3)
669 #endif 513 #endif
670 #undef MANY 514 #undef ANY11M
671 515
672 #define MANY(NAMEANY, COPY_SIMD, COPY_C, BPP, MASK) \ 516 // Any 1 plane. (memset)
673 void NAMEANY(const uint8* src_y, uint8* dst_y, int width) { \ 517 #define ANY1(NAMEANY, ANY_SIMD, T, BPP, MASK) \
674 int n = width & ~MASK; \ 518 void NAMEANY(uint8* dst_ptr, T v32, int width) { \
675 int r = width & MASK; \ 519 SIMD_ALIGNED(uint8 temp[64]); \
676 if (n > 0) { \ 520 int r = width & MASK; \
677 COPY_SIMD(src_y, dst_y, n); \ 521 int n = width & ~MASK; \
678 } \ 522 if (n > 0) { \
679 COPY_C(src_y + n * BPP, dst_y + n * BPP, r); \ 523 ANY_SIMD(dst_ptr, v32, n); \
680 } 524 } \
681 525 ANY_SIMD(temp, v32, MASK + 1); \
682 #ifdef HAS_COPYROW_AVX 526 memcpy(dst_ptr + n * BPP, temp, r * BPP); \
683 MANY(CopyRow_Any_AVX, CopyRow_AVX, CopyRow_C, 1, 63)
684 #endif
685 #ifdef HAS_COPYROW_SSE2
686 MANY(CopyRow_Any_SSE2, CopyRow_SSE2, CopyRow_C, 1, 31)
687 #endif
688 #ifdef HAS_COPYROW_NEON
689 MANY(CopyRow_Any_NEON, CopyRow_NEON, CopyRow_C, 1, 31)
690 #endif
691 #undef MANY
692
693 #define SETANY(NAMEANY, SET_SIMD, SET_C, T, BPP, MASK) \
694 void NAMEANY(uint8* dst_y, T v8, int width) { \
695 int n = width & ~MASK; \
696 int r = width & MASK; \
697 if (n > 0) { \
698 SET_SIMD(dst_y, v8, n); \
699 } \
700 SET_C(dst_y + n * BPP, v8, r); \
701 } 527 }
702 528
703 #ifdef HAS_SETROW_X86 529 #ifdef HAS_SETROW_X86
704 SETANY(SetRow_Any_X86, SetRow_X86, SetRow_ERMS, uint8, 1, 3) 530 ANY1(SetRow_Any_X86, SetRow_X86, uint8, 1, 3)
705 #endif 531 #endif
706 #ifdef HAS_SETROW_NEON 532 #ifdef HAS_SETROW_NEON
707 SETANY(SetRow_Any_NEON, SetRow_NEON, SetRow_C, uint8, 1, 15) 533 ANY1(SetRow_Any_NEON, SetRow_NEON, uint8, 1, 15)
708 #endif 534 #endif
709 #ifdef HAS_ARGBSETROW_NEON 535 #ifdef HAS_ARGBSETROW_NEON
710 SETANY(ARGBSetRow_Any_NEON, ARGBSetRow_NEON, ARGBSetRow_C, uint32, 4, 3) 536 ANY1(ARGBSetRow_Any_NEON, ARGBSetRow_NEON, uint32, 4, 3)
711 #endif 537 #endif
712 #undef SETANY 538 #undef ANY1
539
540 // Any 1 to 2. Outputs UV planes.
541 #define ANY12(NAMEANY, ANY_SIMD, UVSHIFT, BPP, DUVSHIFT, MASK) \
542 void NAMEANY(const uint8* src_ptr, uint8* dst_u, uint8* dst_v, int width) {\
543 SIMD_ALIGNED(uint8 temp[128 * 3]); \
544 memset(temp, 0, 128); /* for msan */ \
545 int r = width & MASK; \
546 int n = width & ~MASK; \
547 if (n > 0) { \
548 ANY_SIMD(src_ptr, dst_u, dst_v, n); \
549 } \
550 memcpy(temp, src_ptr + (n >> UVSHIFT) * BPP, SS(r, UVSHIFT) * BPP); \
551 if ((width & 1) && BPP == 4) { /* repeat last 4 bytes for subsampler */ \
552 memcpy(temp + SS(r, UVSHIFT) * BPP, \
553 temp + SS(r, UVSHIFT) * BPP - BPP, 4); \
554 } \
555 ANY_SIMD(temp, temp + 128, temp + 256, MASK + 1); \
556 memcpy(dst_u + (n >> DUVSHIFT), temp + 128, SS(r, DUVSHIFT)); \
557 memcpy(dst_v + (n >> DUVSHIFT), temp + 256, SS(r, DUVSHIFT)); \
558 }
559
560 #ifdef HAS_SPLITUVROW_SSE2
561 ANY12(SplitUVRow_Any_SSE2, SplitUVRow_SSE2, 0, 2, 0, 15)
562 #endif
563 #ifdef HAS_SPLITUVROW_AVX2
564 ANY12(SplitUVRow_Any_AVX2, SplitUVRow_AVX2, 0, 2, 0, 31)
565 #endif
566 #ifdef HAS_SPLITUVROW_NEON
567 ANY12(SplitUVRow_Any_NEON, SplitUVRow_NEON, 0, 2, 0, 15)
568 #endif
569 #ifdef HAS_SPLITUVROW_MIPS_DSPR2
570 ANY12(SplitUVRow_Any_MIPS_DSPR2, SplitUVRow_MIPS_DSPR2, 0, 2, 0, 15)
571 #endif
572 #ifdef HAS_ARGBTOUV444ROW_SSSE3
573 ANY12(ARGBToUV444Row_Any_SSSE3, ARGBToUV444Row_SSSE3, 0, 4, 0, 15)
574 #endif
575 #ifdef HAS_YUY2TOUV422ROW_AVX2
576 ANY12(YUY2ToUV422Row_Any_AVX2, YUY2ToUV422Row_AVX2, 1, 4, 1, 31)
577 ANY12(UYVYToUV422Row_Any_AVX2, UYVYToUV422Row_AVX2, 1, 4, 1, 31)
578 #endif
579 #ifdef HAS_ARGBTOUV422ROW_SSSE3
580 ANY12(ARGBToUV422Row_Any_SSSE3, ARGBToUV422Row_SSSE3, 0, 4, 1, 15)
581 #endif
582 #ifdef HAS_YUY2TOUV422ROW_SSE2
583 ANY12(YUY2ToUV422Row_Any_SSE2, YUY2ToUV422Row_SSE2, 1, 4, 1, 15)
584 ANY12(UYVYToUV422Row_Any_SSE2, UYVYToUV422Row_SSE2, 1, 4, 1, 15)
585 #endif
586 #ifdef HAS_YUY2TOUV422ROW_NEON
587 ANY12(ARGBToUV444Row_Any_NEON, ARGBToUV444Row_NEON, 0, 4, 0, 7)
588 ANY12(ARGBToUV422Row_Any_NEON, ARGBToUV422Row_NEON, 0, 4, 1, 15)
589 ANY12(ARGBToUV411Row_Any_NEON, ARGBToUV411Row_NEON, 0, 4, 2, 31)
590 ANY12(YUY2ToUV422Row_Any_NEON, YUY2ToUV422Row_NEON, 1, 4, 1, 15)
591 ANY12(UYVYToUV422Row_Any_NEON, UYVYToUV422Row_NEON, 1, 4, 1, 15)
592 #endif
593 #undef ANY12
594
595 // Any 1 to 2 with source stride (2 rows of source). Outputs UV planes.
596 // 128 byte row allows for 32 avx ARGB pixels.
597 #define ANY12S(NAMEANY, ANY_SIMD, UVSHIFT, BPP, MASK) \
598 void NAMEANY(const uint8* src_ptr, int src_stride_ptr, \
599 uint8* dst_u, uint8* dst_v, int width) { \
600 SIMD_ALIGNED(uint8 temp[128 * 4]); \
601 memset(temp, 0, 128 * 2); /* for msan */ \
602 int r = width & MASK; \
603 int n = width & ~MASK; \
604 if (n > 0) { \
605 ANY_SIMD(src_ptr, src_stride_ptr, dst_u, dst_v, n); \
606 } \
607 memcpy(temp, src_ptr + (n >> UVSHIFT) * BPP, SS(r, UVSHIFT) * BPP); \
608 memcpy(temp + 128, src_ptr + src_stride_ptr + (n >> UVSHIFT) * BPP, \
609 SS(r, UVSHIFT) * BPP); \
610 if ((width & 1) && BPP == 4) { /* repeat last 4 bytes for subsampler */ \
611 memcpy(temp + SS(r, UVSHIFT) * BPP, \
612 temp + SS(r, UVSHIFT) * BPP - BPP, 4); \
613 memcpy(temp + 128 + SS(r, UVSHIFT) * BPP, \
614 temp + 128 + SS(r, UVSHIFT) * BPP - BPP, 4); \
615 } \
616 ANY_SIMD(temp, 128, temp + 256, temp + 384, MASK + 1); \
617 memcpy(dst_u + (n >> 1), temp + 256, SS(r, 1)); \
618 memcpy(dst_v + (n >> 1), temp + 384, SS(r, 1)); \
619 }
620
621 #ifdef HAS_ARGBTOUVROW_AVX2
622 ANY12S(ARGBToUVRow_Any_AVX2, ARGBToUVRow_AVX2, 0, 4, 31)
623 #endif
624 #ifdef HAS_ARGBTOUVROW_SSSE3
625 ANY12S(ARGBToUVRow_Any_SSSE3, ARGBToUVRow_SSSE3, 0, 4, 15)
626 ANY12S(ARGBToUVJRow_Any_SSSE3, ARGBToUVJRow_SSSE3, 0, 4, 15)
627 ANY12S(BGRAToUVRow_Any_SSSE3, BGRAToUVRow_SSSE3, 0, 4, 15)
628 ANY12S(ABGRToUVRow_Any_SSSE3, ABGRToUVRow_SSSE3, 0, 4, 15)
629 ANY12S(RGBAToUVRow_Any_SSSE3, RGBAToUVRow_SSSE3, 0, 4, 15)
630 #endif
631 #ifdef HAS_YUY2TOUVROW_AVX2
632 ANY12S(YUY2ToUVRow_Any_AVX2, YUY2ToUVRow_AVX2, 1, 4, 31)
633 ANY12S(UYVYToUVRow_Any_AVX2, UYVYToUVRow_AVX2, 1, 4, 31)
634 #endif
635 #ifdef HAS_YUY2TOUVROW_SSE2
636 ANY12S(YUY2ToUVRow_Any_SSE2, YUY2ToUVRow_SSE2, 1, 4, 15)
637 ANY12S(UYVYToUVRow_Any_SSE2, UYVYToUVRow_SSE2, 1, 4, 15)
638 #endif
639 #ifdef HAS_ARGBTOUVROW_NEON
640 ANY12S(ARGBToUVRow_Any_NEON, ARGBToUVRow_NEON, 0, 4, 15)
641 #endif
642 #ifdef HAS_ARGBTOUVJROW_NEON
643 ANY12S(ARGBToUVJRow_Any_NEON, ARGBToUVJRow_NEON, 0, 4, 15)
644 #endif
645 #ifdef HAS_BGRATOUVROW_NEON
646 ANY12S(BGRAToUVRow_Any_NEON, BGRAToUVRow_NEON, 0, 4, 15)
647 #endif
648 #ifdef HAS_ABGRTOUVROW_NEON
649 ANY12S(ABGRToUVRow_Any_NEON, ABGRToUVRow_NEON, 0, 4, 15)
650 #endif
651 #ifdef HAS_RGBATOUVROW_NEON
652 ANY12S(RGBAToUVRow_Any_NEON, RGBAToUVRow_NEON, 0, 4, 15)
653 #endif
654 #ifdef HAS_RGB24TOUVROW_NEON
655 ANY12S(RGB24ToUVRow_Any_NEON, RGB24ToUVRow_NEON, 0, 3, 15)
656 #endif
657 #ifdef HAS_RAWTOUVROW_NEON
658 ANY12S(RAWToUVRow_Any_NEON, RAWToUVRow_NEON, 0, 3, 15)
659 #endif
660 #ifdef HAS_RGB565TOUVROW_NEON
661 ANY12S(RGB565ToUVRow_Any_NEON, RGB565ToUVRow_NEON, 0, 2, 15)
662 #endif
663 #ifdef HAS_ARGB1555TOUVROW_NEON
664 ANY12S(ARGB1555ToUVRow_Any_NEON, ARGB1555ToUVRow_NEON, 0, 2, 15)
665 #endif
666 #ifdef HAS_ARGB4444TOUVROW_NEON
667 ANY12S(ARGB4444ToUVRow_Any_NEON, ARGB4444ToUVRow_NEON, 0, 2, 15)
668 #endif
669 #ifdef HAS_YUY2TOUVROW_NEON
670 ANY12S(YUY2ToUVRow_Any_NEON, YUY2ToUVRow_NEON, 1, 4, 15)
671 #endif
672 #ifdef HAS_UYVYTOUVROW_NEON
673 ANY12S(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, 1, 4, 15)
674 #endif
675 #undef ANY12S
713 676
714 #ifdef __cplusplus 677 #ifdef __cplusplus
715 } // extern "C" 678 } // extern "C"
716 } // namespace libyuv 679 } // namespace libyuv
717 #endif 680 #endif
OLDNEW
« no previous file with comments | « source/libvpx/third_party/libyuv/source/rotate_win.cc ('k') | source/libvpx/third_party/libyuv/source/row_common.cc » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698