react-native-vision-camera-spoof-detector 1.0.22 → 1.0.23

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (193) hide show
  1. package/README.md +442 -442
  2. package/android/build.gradle +12 -4
  3. package/android/src/main/cpp/CMakeLists.txt +101 -94
  4. package/android/src/main/cpp/libyuv/.clang-format +6 -6
  5. package/android/src/main/cpp/libyuv/.gn +40 -40
  6. package/android/src/main/cpp/libyuv/.vpython3 +410 -410
  7. package/android/src/main/cpp/libyuv/AUTHORS +7 -7
  8. package/android/src/main/cpp/libyuv/Android.bp +202 -202
  9. package/android/src/main/cpp/libyuv/Android.mk +106 -106
  10. package/android/src/main/cpp/libyuv/BUILD.gn +490 -490
  11. package/android/src/main/cpp/libyuv/CM_linux_packages.cmake +70 -70
  12. package/android/src/main/cpp/libyuv/CMakeLists.txt +259 -252
  13. package/android/src/main/cpp/libyuv/DEPS +935 -935
  14. package/android/src/main/cpp/libyuv/DIR_METADATA +3 -3
  15. package/android/src/main/cpp/libyuv/LICENSE +29 -29
  16. package/android/src/main/cpp/libyuv/OWNERS +11 -11
  17. package/android/src/main/cpp/libyuv/PATENTS +23 -23
  18. package/android/src/main/cpp/libyuv/PRESUBMIT.py +65 -65
  19. package/android/src/main/cpp/libyuv/README.chromium +11 -11
  20. package/android/src/main/cpp/libyuv/README.md +19 -19
  21. package/android/src/main/cpp/libyuv/build_overrides/build.gni +60 -60
  22. package/android/src/main/cpp/libyuv/build_overrides/gtest.gni +19 -19
  23. package/android/src/main/cpp/libyuv/build_overrides/partition_alloc.gni +17 -17
  24. package/android/src/main/cpp/libyuv/codereview.settings +5 -5
  25. package/android/src/main/cpp/libyuv/docs/deprecated_builds.md +409 -409
  26. package/android/src/main/cpp/libyuv/docs/environment_variables.md +64 -64
  27. package/android/src/main/cpp/libyuv/docs/feature_detection.md +108 -108
  28. package/android/src/main/cpp/libyuv/docs/filtering.md +196 -196
  29. package/android/src/main/cpp/libyuv/docs/formats.md +208 -208
  30. package/android/src/main/cpp/libyuv/docs/getting_started.md +296 -296
  31. package/android/src/main/cpp/libyuv/docs/rotation.md +107 -107
  32. package/android/src/main/cpp/libyuv/download_vs_toolchain.py +29 -29
  33. package/android/src/main/cpp/libyuv/include/libyuv/basic_types.h +68 -68
  34. package/android/src/main/cpp/libyuv/include/libyuv/compare.h +111 -111
  35. package/android/src/main/cpp/libyuv/include/libyuv/compare_row.h +112 -112
  36. package/android/src/main/cpp/libyuv/include/libyuv/convert.h +1117 -1117
  37. package/android/src/main/cpp/libyuv/include/libyuv/convert_argb.h +2335 -2335
  38. package/android/src/main/cpp/libyuv/include/libyuv/convert_from.h +203 -203
  39. package/android/src/main/cpp/libyuv/include/libyuv/convert_from_argb.h +394 -394
  40. package/android/src/main/cpp/libyuv/include/libyuv/cpu_id.h +147 -147
  41. package/android/src/main/cpp/libyuv/include/libyuv/cpu_support.h +99 -99
  42. package/android/src/main/cpp/libyuv/include/libyuv/loongson_intrinsics.h +1949 -1949
  43. package/android/src/main/cpp/libyuv/include/libyuv/mjpeg_decoder.h +195 -195
  44. package/android/src/main/cpp/libyuv/include/libyuv/planar_functions.h +1131 -1131
  45. package/android/src/main/cpp/libyuv/include/libyuv/rotate.h +296 -296
  46. package/android/src/main/cpp/libyuv/include/libyuv/rotate_argb.h +37 -37
  47. package/android/src/main/cpp/libyuv/include/libyuv/rotate_row.h +265 -265
  48. package/android/src/main/cpp/libyuv/include/libyuv/row.h +6738 -6738
  49. package/android/src/main/cpp/libyuv/include/libyuv/row_sve.h +2154 -2154
  50. package/android/src/main/cpp/libyuv/include/libyuv/scale.h +336 -336
  51. package/android/src/main/cpp/libyuv/include/libyuv/scale_argb.h +76 -76
  52. package/android/src/main/cpp/libyuv/include/libyuv/scale_rgb.h +42 -42
  53. package/android/src/main/cpp/libyuv/include/libyuv/scale_row.h +1726 -1726
  54. package/android/src/main/cpp/libyuv/include/libyuv/scale_uv.h +51 -51
  55. package/android/src/main/cpp/libyuv/include/libyuv/version.h +16 -16
  56. package/android/src/main/cpp/libyuv/include/libyuv/video_common.h +222 -222
  57. package/android/src/main/cpp/libyuv/include/libyuv.h +33 -33
  58. package/android/src/main/cpp/libyuv/infra/config/OWNERS +3 -3
  59. package/android/src/main/cpp/libyuv/infra/config/PRESUBMIT.py +17 -17
  60. package/android/src/main/cpp/libyuv/infra/config/README.md +2 -2
  61. package/android/src/main/cpp/libyuv/infra/config/codereview.settings +6 -6
  62. package/android/src/main/cpp/libyuv/infra/config/commit-queue.cfg +144 -144
  63. package/android/src/main/cpp/libyuv/infra/config/cr-buildbucket.cfg +1185 -1185
  64. package/android/src/main/cpp/libyuv/infra/config/luci-logdog.cfg +9 -9
  65. package/android/src/main/cpp/libyuv/infra/config/luci-milo.cfg +246 -246
  66. package/android/src/main/cpp/libyuv/infra/config/luci-scheduler.cfg +385 -385
  67. package/android/src/main/cpp/libyuv/infra/config/main.star +402 -402
  68. package/android/src/main/cpp/libyuv/infra/config/project.cfg +16 -16
  69. package/android/src/main/cpp/libyuv/infra/config/realms.cfg +111 -111
  70. package/android/src/main/cpp/libyuv/libyuv.gni +34 -34
  71. package/android/src/main/cpp/libyuv/libyuv.gyp +149 -149
  72. package/android/src/main/cpp/libyuv/libyuv.gypi +87 -87
  73. package/android/src/main/cpp/libyuv/linux.mk +96 -96
  74. package/android/src/main/cpp/libyuv/public.mk +13 -13
  75. package/android/src/main/cpp/libyuv/pylintrc +49 -49
  76. package/android/src/main/cpp/libyuv/riscv_script/prepare_toolchain_qemu.sh +74 -74
  77. package/android/src/main/cpp/libyuv/riscv_script/riscv-clang.cmake +56 -56
  78. package/android/src/main/cpp/libyuv/riscv_script/run_qemu.sh +15 -15
  79. package/android/src/main/cpp/libyuv/source/compare.cc +435 -435
  80. package/android/src/main/cpp/libyuv/source/compare_common.cc +74 -74
  81. package/android/src/main/cpp/libyuv/source/compare_gcc.cc +362 -362
  82. package/android/src/main/cpp/libyuv/source/compare_neon.cc +96 -96
  83. package/android/src/main/cpp/libyuv/source/compare_neon64.cc +223 -223
  84. package/android/src/main/cpp/libyuv/source/compare_win.cc +241 -241
  85. package/android/src/main/cpp/libyuv/source/convert.cc +4746 -4746
  86. package/android/src/main/cpp/libyuv/source/convert_argb.cc +9179 -9179
  87. package/android/src/main/cpp/libyuv/source/convert_from.cc +866 -866
  88. package/android/src/main/cpp/libyuv/source/convert_from_argb.cc +3671 -3671
  89. package/android/src/main/cpp/libyuv/source/convert_jpeg.cc +602 -602
  90. package/android/src/main/cpp/libyuv/source/convert_to_argb.cc +391 -391
  91. package/android/src/main/cpp/libyuv/source/convert_to_i420.cc +288 -288
  92. package/android/src/main/cpp/libyuv/source/cpu_id.cc +496 -496
  93. package/android/src/main/cpp/libyuv/source/mjpeg_decoder.cc +580 -580
  94. package/android/src/main/cpp/libyuv/source/mjpeg_validate.cc +71 -71
  95. package/android/src/main/cpp/libyuv/source/planar_functions.cc +5663 -5663
  96. package/android/src/main/cpp/libyuv/source/rotate.cc +1241 -1241
  97. package/android/src/main/cpp/libyuv/source/rotate_any.cc +76 -76
  98. package/android/src/main/cpp/libyuv/source/rotate_argb.cc +259 -259
  99. package/android/src/main/cpp/libyuv/source/rotate_common.cc +208 -208
  100. package/android/src/main/cpp/libyuv/source/rotate_gcc.cc +505 -505
  101. package/android/src/main/cpp/libyuv/source/rotate_lsx.cc +233 -233
  102. package/android/src/main/cpp/libyuv/source/rotate_neon.cc +219 -219
  103. package/android/src/main/cpp/libyuv/source/rotate_neon64.cc +273 -273
  104. package/android/src/main/cpp/libyuv/source/rotate_sme.cc +174 -174
  105. package/android/src/main/cpp/libyuv/source/rotate_win.cc +253 -253
  106. package/android/src/main/cpp/libyuv/source/row_any.cc +2519 -2519
  107. package/android/src/main/cpp/libyuv/source/row_common.cc +4461 -4461
  108. package/android/src/main/cpp/libyuv/source/row_gcc.cc +9570 -9570
  109. package/android/src/main/cpp/libyuv/source/row_lasx.cc +2343 -2343
  110. package/android/src/main/cpp/libyuv/source/row_lsx.cc +3030 -3030
  111. package/android/src/main/cpp/libyuv/source/row_neon.cc +4026 -4026
  112. package/android/src/main/cpp/libyuv/source/row_neon64.cc +5617 -5617
  113. package/android/src/main/cpp/libyuv/source/row_rvv.cc +2599 -2599
  114. package/android/src/main/cpp/libyuv/source/row_sme.cc +1183 -1183
  115. package/android/src/main/cpp/libyuv/source/row_sve.cc +1088 -1088
  116. package/android/src/main/cpp/libyuv/source/row_win.cc +6453 -6453
  117. package/android/src/main/cpp/libyuv/source/scale.cc +2710 -2710
  118. package/android/src/main/cpp/libyuv/source/scale_any.cc +991 -991
  119. package/android/src/main/cpp/libyuv/source/scale_argb.cc +1158 -1158
  120. package/android/src/main/cpp/libyuv/source/scale_common.cc +1977 -1977
  121. package/android/src/main/cpp/libyuv/source/scale_gcc.cc +2947 -2947
  122. package/android/src/main/cpp/libyuv/source/scale_lsx.cc +739 -739
  123. package/android/src/main/cpp/libyuv/source/scale_neon.cc +1449 -1449
  124. package/android/src/main/cpp/libyuv/source/scale_neon64.cc +1552 -1552
  125. package/android/src/main/cpp/libyuv/source/scale_rgb.cc +82 -82
  126. package/android/src/main/cpp/libyuv/source/scale_rvv.cc +1971 -1971
  127. package/android/src/main/cpp/libyuv/source/scale_sme.cc +555 -555
  128. package/android/src/main/cpp/libyuv/source/scale_uv.cc +1159 -1159
  129. package/android/src/main/cpp/libyuv/source/scale_win.cc +1392 -1392
  130. package/android/src/main/cpp/libyuv/source/test.sh +35 -35
  131. package/android/src/main/cpp/libyuv/source/video_common.cc +62 -62
  132. package/android/src/main/cpp/libyuv/tools_libyuv/OWNERS +4 -4
  133. package/android/src/main/cpp/libyuv/tools_libyuv/autoroller/roll_deps.py +931 -931
  134. package/android/src/main/cpp/libyuv/tools_libyuv/autoroller/unittests/roll_deps_test.py +164 -164
  135. package/android/src/main/cpp/libyuv/tools_libyuv/autoroller/unittests/testdata/DEPS +21 -21
  136. package/android/src/main/cpp/libyuv/tools_libyuv/autoroller/unittests/testdata/DEPS.chromium.new +13 -13
  137. package/android/src/main/cpp/libyuv/tools_libyuv/autoroller/unittests/testdata/DEPS.chromium.old +13 -13
  138. package/android/src/main/cpp/libyuv/tools_libyuv/get_landmines.py +38 -38
  139. package/android/src/main/cpp/libyuv/tools_libyuv/msan/OWNERS +3 -3
  140. package/android/src/main/cpp/libyuv/tools_libyuv/msan/blacklist.txt +9 -9
  141. package/android/src/main/cpp/libyuv/tools_libyuv/ubsan/OWNERS +3 -3
  142. package/android/src/main/cpp/libyuv/tools_libyuv/ubsan/blacklist.txt +15 -15
  143. package/android/src/main/cpp/libyuv/tools_libyuv/ubsan/vptr_blacklist.txt +25 -25
  144. package/android/src/main/cpp/libyuv/unit_test/basictypes_test.cc +43 -43
  145. package/android/src/main/cpp/libyuv/unit_test/color_test.cc +848 -848
  146. package/android/src/main/cpp/libyuv/unit_test/compare_test.cc +739 -739
  147. package/android/src/main/cpp/libyuv/unit_test/convert_argb_test.cc +2867 -2867
  148. package/android/src/main/cpp/libyuv/unit_test/convert_test.cc +2133 -2133
  149. package/android/src/main/cpp/libyuv/unit_test/cpu_test.cc +427 -427
  150. package/android/src/main/cpp/libyuv/unit_test/cpu_thread_test.cc +63 -63
  151. package/android/src/main/cpp/libyuv/unit_test/math_test.cc +160 -160
  152. package/android/src/main/cpp/libyuv/unit_test/planar_test.cc +4731 -4731
  153. package/android/src/main/cpp/libyuv/unit_test/rotate_argb_test.cc +334 -334
  154. package/android/src/main/cpp/libyuv/unit_test/rotate_test.cc +962 -962
  155. package/android/src/main/cpp/libyuv/unit_test/scale_argb_test.cc +590 -590
  156. package/android/src/main/cpp/libyuv/unit_test/scale_plane_test.cc +465 -465
  157. package/android/src/main/cpp/libyuv/unit_test/scale_rgb_test.cc +280 -280
  158. package/android/src/main/cpp/libyuv/unit_test/scale_test.cc +1135 -1135
  159. package/android/src/main/cpp/libyuv/unit_test/scale_uv_test.cc +249 -249
  160. package/android/src/main/cpp/libyuv/unit_test/testdata/arm_v7.txt +12 -12
  161. package/android/src/main/cpp/libyuv/unit_test/testdata/mips.txt +7 -7
  162. package/android/src/main/cpp/libyuv/unit_test/testdata/mips_loongson2k.txt +5 -5
  163. package/android/src/main/cpp/libyuv/unit_test/testdata/mips_loongson3.txt +10 -10
  164. package/android/src/main/cpp/libyuv/unit_test/testdata/mips_loongson_mmi.txt +7 -7
  165. package/android/src/main/cpp/libyuv/unit_test/testdata/riscv64.txt +3 -3
  166. package/android/src/main/cpp/libyuv/unit_test/testdata/riscv64_rvv.txt +3 -3
  167. package/android/src/main/cpp/libyuv/unit_test/testdata/riscv64_rvv_zvfh.txt +3 -3
  168. package/android/src/main/cpp/libyuv/unit_test/testdata/tegra3.txt +23 -23
  169. package/android/src/main/cpp/libyuv/unit_test/unit_test.cc +581 -581
  170. package/android/src/main/cpp/libyuv/unit_test/unit_test.h +232 -232
  171. package/android/src/main/cpp/libyuv/unit_test/video_common_test.cc +112 -112
  172. package/android/src/main/cpp/libyuv/util/Makefile +9 -9
  173. package/android/src/main/cpp/libyuv/util/color.cc +120 -120
  174. package/android/src/main/cpp/libyuv/util/compare.cc +67 -67
  175. package/android/src/main/cpp/libyuv/util/cpuid.c +211 -211
  176. package/android/src/main/cpp/libyuv/util/i444tonv12_eg.cc +27 -27
  177. package/android/src/main/cpp/libyuv/util/psnr.cc +291 -291
  178. package/android/src/main/cpp/libyuv/util/psnr.h +47 -47
  179. package/android/src/main/cpp/libyuv/util/psnr_main.cc +620 -620
  180. package/android/src/main/cpp/libyuv/util/ssim.cc +364 -364
  181. package/android/src/main/cpp/libyuv/util/ssim.h +38 -38
  182. package/android/src/main/cpp/libyuv/util/yuvconstants.c +114 -114
  183. package/android/src/main/cpp/libyuv/util/yuvconvert.cc +367 -367
  184. package/android/src/main/cpp/libyuv/winarm.mk +47 -47
  185. package/package.json +86 -86
  186. package/android/.gradle/8.9/checksums/checksums.lock +0 -0
  187. package/android/.gradle/8.9/dependencies-accessors/gc.properties +0 -0
  188. package/android/.gradle/8.9/fileChanges/last-build.bin +0 -0
  189. package/android/.gradle/8.9/fileHashes/fileHashes.lock +0 -0
  190. package/android/.gradle/8.9/gc.properties +0 -0
  191. package/android/.gradle/buildOutputCleanup/buildOutputCleanup.lock +0 -0
  192. package/android/.gradle/buildOutputCleanup/cache.properties +0 -2
  193. package/android/.gradle/vcs-1/gc.properties +0 -0
package/android/src/main/cpp/libyuv/source/scale_sme.cc CHANGED
@@ -1,555 +1,555 @@
1
- /*
2
- * Copyright 2024 The LibYuv Project Authors. All rights reserved.
3
- *
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
6
- * tree. An additional intellectual property rights grant can be found
7
- * in the file PATENTS. All contributing project authors may
8
- * be found in the AUTHORS file in the root of the source tree.
9
- */
10
-
11
- #include "libyuv/scale_row.h"
12
-
13
- #ifdef __cplusplus
14
- namespace libyuv {
15
- extern "C" {
16
- #endif
17
-
18
- #if !defined(LIBYUV_DISABLE_SME) && defined(CLANG_HAS_SME) && \
19
- defined(__aarch64__)
20
-
21
- __arm_locally_streaming void ScaleRowDown2_SME(const uint8_t* src_ptr,
22
- ptrdiff_t src_stride,
23
- uint8_t* dst,
24
- int dst_width) {
25
- // Streaming-SVE only, no use of ZA tile.
26
- (void)src_stride;
27
- int vl;
28
- asm volatile(
29
- "cntb %x[vl] \n"
30
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
31
- "b.lt 2f \n"
32
-
33
- "1: \n"
34
- "ptrue p0.b \n"
35
- "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
36
- "incb %[src_ptr], all, mul #2 \n"
37
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
38
- "st1b {z1.b}, p0, [%[dst_ptr]] \n"
39
- "incb %[dst_ptr] \n"
40
- "b.ge 1b \n"
41
-
42
- "2: \n"
43
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
44
- "b.eq 99f \n"
45
-
46
- "whilelt p0.b, wzr, %w[dst_width] \n"
47
- "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
48
- "st1b {z1.b}, p0, [%[dst_ptr]] \n"
49
-
50
- "99: \n"
51
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
52
- [dst_ptr] "+r"(dst), // %[dst_ptr]
53
- [dst_width] "+r"(dst_width), // %[dst_width]
54
- [vl] "=r"(vl) // %[vl]
55
- :
56
- : "memory", "cc", "z0", "z1", "p0");
57
- }
58
-
59
- __arm_locally_streaming void ScaleRowDown2_16_SME(const uint16_t* src_ptr,
60
- ptrdiff_t src_stride,
61
- uint16_t* dst,
62
- int dst_width) {
63
- // Streaming-SVE only, no use of ZA tile.
64
- (void)src_stride;
65
- int vl;
66
- asm volatile(
67
- "cnth %x[vl] \n"
68
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
69
- "b.lt 2f \n"
70
-
71
- "1: \n"
72
- "ptrue p0.h \n"
73
- "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
74
- "incb %[src_ptr], all, mul #2 \n"
75
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
76
- "st1h {z1.h}, p0, [%[dst_ptr]] \n"
77
- "incb %[dst_ptr] \n"
78
- "b.ge 1b \n"
79
-
80
- "2: \n"
81
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
82
- "b.eq 99f \n"
83
-
84
- "whilelt p0.h, wzr, %w[dst_width] \n"
85
- "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
86
- "st1h {z1.h}, p0, [%[dst_ptr]] \n"
87
-
88
- "99: \n"
89
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
90
- [dst_ptr] "+r"(dst), // %[dst_ptr]
91
- [dst_width] "+r"(dst_width), // %[dst_width]
92
- [vl] "=r"(vl) // %[vl]
93
- :
94
- : "memory", "cc", "z0", "z1", "p0");
95
- }
96
-
97
- __arm_locally_streaming void ScaleRowDown2Linear_SME(const uint8_t* src_ptr,
98
- ptrdiff_t src_stride,
99
- uint8_t* dst,
100
- int dst_width) {
101
- // Streaming-SVE only, no use of ZA tile.
102
- (void)src_stride;
103
- int vl;
104
- asm volatile(
105
- "cntb %x[vl] \n"
106
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
107
- "b.lt 2f \n"
108
-
109
- "1: \n"
110
- "ptrue p0.b \n"
111
- "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
112
- "incb %[src_ptr], all, mul #2 \n"
113
- "urhadd z0.b, p0/m, z0.b, z1.b \n"
114
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
115
- "st1b {z0.b}, p0, [%[dst_ptr]] \n"
116
- "incb %[dst_ptr] \n"
117
- "b.ge 1b \n"
118
-
119
- "2: \n"
120
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
121
- "b.eq 99f \n"
122
-
123
- "whilelt p0.b, wzr, %w[dst_width] \n"
124
- "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
125
- "urhadd z0.b, p0/m, z0.b, z1.b \n"
126
- "st1b {z0.b}, p0, [%[dst_ptr]] \n"
127
-
128
- "99: \n"
129
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
130
- [dst_ptr] "+r"(dst), // %[dst_ptr]
131
- [dst_width] "+r"(dst_width), // %[dst_width]
132
- [vl] "=r"(vl) // %[vl]
133
- :
134
- : "memory", "cc", "z0", "z1", "p0");
135
- }
136
-
137
- __arm_locally_streaming void ScaleRowDown2Linear_16_SME(const uint16_t* src_ptr,
138
- ptrdiff_t src_stride,
139
- uint16_t* dst,
140
- int dst_width) {
141
- // Streaming-SVE only, no use of ZA tile.
142
- (void)src_stride;
143
- int vl;
144
- asm volatile(
145
- "cnth %x[vl] \n"
146
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
147
- "b.lt 2f \n"
148
-
149
- "1: \n"
150
- "ptrue p0.h \n"
151
- "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
152
- "incb %[src_ptr], all, mul #2 \n"
153
- "urhadd z0.h, p0/m, z0.h, z1.h \n"
154
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
155
- "st1h {z0.h}, p0, [%[dst_ptr]] \n"
156
- "incb %[dst_ptr] \n"
157
- "b.ge 1b \n"
158
-
159
- "2: \n"
160
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
161
- "b.eq 99f \n"
162
-
163
- "whilelt p0.h, wzr, %w[dst_width] \n"
164
- "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
165
- "urhadd z0.h, p0/m, z0.h, z1.h \n"
166
- "st1h {z0.h}, p0, [%[dst_ptr]] \n"
167
-
168
- "99: \n"
169
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
170
- [dst_ptr] "+r"(dst), // %[dst_ptr]
171
- [dst_width] "+r"(dst_width), // %[dst_width]
172
- [vl] "=r"(vl) // %[vl]
173
- :
174
- : "memory", "cc", "z0", "z1", "p0");
175
- }
176
-
177
- #define SCALEROWDOWN2BOX_SVE \
178
- "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n" \
179
- "ld2b {z2.b, z3.b}, p0/z, [%[src2_ptr]] \n" \
180
- "incb %[src_ptr], all, mul #2 \n" \
181
- "incb %[src2_ptr], all, mul #2 \n" \
182
- "uaddlb z4.h, z0.b, z1.b \n" \
183
- "uaddlt z5.h, z0.b, z1.b \n" \
184
- "uaddlb z6.h, z2.b, z3.b \n" \
185
- "uaddlt z7.h, z2.b, z3.b \n" \
186
- "add z4.h, z4.h, z6.h \n" \
187
- "add z5.h, z5.h, z7.h \n" \
188
- "rshrnb z0.b, z4.h, #2 \n" \
189
- "rshrnt z0.b, z5.h, #2 \n" \
190
- "subs %w[dst_width], %w[dst_width], %w[vl] \n" \
191
- "st1b {z0.b}, p0, [%[dst_ptr]] \n" \
192
- "incb %[dst_ptr] \n"
193
-
194
- __arm_locally_streaming void ScaleRowDown2Box_SME(const uint8_t* src_ptr,
195
- ptrdiff_t src_stride,
196
- uint8_t* dst,
197
- int dst_width) {
198
- // Streaming-SVE only, no use of ZA tile.
199
- const uint8_t* src2_ptr = src_ptr + src_stride;
200
- int vl;
201
- asm volatile(
202
- "cntb %x[vl] \n"
203
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
204
- "b.lt 2f \n"
205
-
206
- "ptrue p0.b \n"
207
- "1: \n" //
208
- SCALEROWDOWN2BOX_SVE
209
- "b.ge 1b \n"
210
-
211
- "2: \n"
212
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
213
- "b.eq 99f \n"
214
-
215
- "whilelt p0.b, wzr, %w[dst_width] \n" //
216
- SCALEROWDOWN2BOX_SVE
217
-
218
- "99: \n"
219
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
220
- [src2_ptr] "+r"(src2_ptr), // %[src2_ptr]
221
- [dst_ptr] "+r"(dst), // %[dst_ptr]
222
- [dst_width] "+r"(dst_width), // %[dst_width]
223
- [vl] "=r"(vl) // %[vl]
224
- :
225
- : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0");
226
- }
227
-
228
- #undef SCALEROWDOWN2BOX_SVE
229
-
230
- #define SCALEROWDOWN2BOX_16_SVE \
231
- "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n" \
232
- "ld2h {z2.h, z3.h}, p0/z, [%[src2_ptr]] \n" \
233
- "incb %[src_ptr], all, mul #2 \n" \
234
- "incb %[src2_ptr], all, mul #2 \n" \
235
- "uaddlb z4.s, z0.h, z1.h \n" \
236
- "uaddlt z5.s, z0.h, z1.h \n" \
237
- "uaddlb z6.s, z2.h, z3.h \n" \
238
- "uaddlt z7.s, z2.h, z3.h \n" \
239
- "add z4.s, z4.s, z6.s \n" \
240
- "add z5.s, z5.s, z7.s \n" \
241
- "rshrnb z0.h, z4.s, #2 \n" \
242
- "rshrnt z0.h, z5.s, #2 \n" \
243
- "subs %w[dst_width], %w[dst_width], %w[vl] \n" \
244
- "st1h {z0.h}, p0, [%[dst_ptr]] \n" \
245
- "incb %[dst_ptr] \n"
246
-
247
- __arm_locally_streaming void ScaleRowDown2Box_16_SME(const uint16_t* src_ptr,
248
- ptrdiff_t src_stride,
249
- uint16_t* dst,
250
- int dst_width) {
251
- // Streaming-SVE only, no use of ZA tile.
252
- const uint16_t* src2_ptr = src_ptr + src_stride;
253
- int vl;
254
- asm volatile(
255
- "cnth %x[vl] \n"
256
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
257
- "b.lt 2f \n"
258
-
259
- "ptrue p0.h \n"
260
- "1: \n" //
261
- SCALEROWDOWN2BOX_16_SVE
262
- "b.ge 1b \n"
263
-
264
- "2: \n"
265
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
266
- "b.eq 99f \n"
267
-
268
- "whilelt p0.h, wzr, %w[dst_width] \n" //
269
- SCALEROWDOWN2BOX_16_SVE
270
-
271
- "99: \n"
272
- : [src_ptr] "+r"(src_ptr), // %[src_ptr]
273
- [src2_ptr] "+r"(src2_ptr), // %[src2_ptr]
274
- [dst_ptr] "+r"(dst), // %[dst_ptr]
275
- [dst_width] "+r"(dst_width), // %[dst_width]
276
- [vl] "=r"(vl) // %[vl]
277
- :
278
- : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0");
279
- }
280
-
281
- #undef SCALEROWDOWN2BOX_16_SVE
282
-
283
- __arm_locally_streaming void ScaleUVRowDown2_SME(const uint8_t* src_uv,
284
- ptrdiff_t src_stride,
285
- uint8_t* dst_uv,
286
- int dst_width) {
287
- // Streaming-SVE only, no use of ZA tile.
288
- (void)src_stride;
289
- int vl;
290
- asm volatile(
291
- "cnth %x[vl] \n"
292
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
293
- "b.lt 2f \n"
294
-
295
- "1: \n"
296
- "ptrue p0.b \n"
297
- "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
298
- "incb %[src_uv], all, mul #2 \n"
299
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
300
- "st1h {z1.h}, p0, [%[dst_uv]] \n"
301
- "incb %[dst_uv] \n"
302
- "b.ge 1b \n"
303
-
304
- "2: \n"
305
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
306
- "b.eq 99f \n"
307
-
308
- "whilelt p0.h, wzr, %w[dst_width] \n"
309
- "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
310
- "st1h {z1.h}, p0, [%[dst_uv]] \n"
311
-
312
- "99: \n"
313
- : [src_uv] "+r"(src_uv), // %[src_uv]
314
- [dst_uv] "+r"(dst_uv), // %[dst_uv]
315
- [dst_width] "+r"(dst_width), // %[dst_width]
316
- [vl] "=r"(vl) // %[vl]
317
- :
318
- : "memory", "cc", "z0", "z1", "p0");
319
- }
320
-
321
- __arm_locally_streaming void ScaleUVRowDown2Linear_SME(const uint8_t* src_uv,
322
- ptrdiff_t src_stride,
323
- uint8_t* dst_uv,
324
- int dst_width) {
325
- // Streaming-SVE only, no use of ZA tile.
326
- (void)src_stride;
327
- int vl;
328
- asm volatile(
329
- "cnth %x[vl] \n"
330
- "ptrue p1.b \n"
331
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
332
- "b.lt 2f \n"
333
-
334
- "ptrue p0.h \n"
335
- "1: \n"
336
- "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
337
- "incb %[src_uv], all, mul #2 \n"
338
- "urhadd z0.b, p1/m, z0.b, z1.b \n"
339
- "st1h {z0.h}, p0, [%[dst_uv]] \n"
340
- "incb %[dst_uv], all, mul #1 \n"
341
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
342
- "b.ge 1b \n"
343
-
344
- "2: \n"
345
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
346
- "b.eq 99f \n"
347
-
348
- "whilelt p0.h, wzr, %w[dst_width] \n"
349
- "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
350
- "urhadd z0.b, p1/m, z0.b, z1.b \n"
351
- "st1h {z0.h}, p0, [%[dst_uv]] \n"
352
-
353
- "99: \n"
354
- : [src_uv] "+r"(src_uv), // %[src_uv]
355
- [dst_uv] "+r"(dst_uv), // %[dst_uv]
356
- [dst_width] "+r"(dst_width), // %[dst_width]
357
- [vl] "=r"(vl) // %[vl]
358
- :
359
- : "z0", "z1", "p0", "p1");
360
- }
361
-
362
- #define SCALEUVROWDOWN2BOX_SVE \
363
- "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n" \
364
- "ld2h {z2.h, z3.h}, p0/z, [%[src2_uv]] \n" \
365
- "incb %[src_uv], all, mul #2 \n" \
366
- "incb %[src2_uv], all, mul #2 \n" \
367
- "uaddlb z4.h, z0.b, z1.b \n" \
368
- "uaddlt z5.h, z0.b, z1.b \n" \
369
- "uaddlb z6.h, z2.b, z3.b \n" \
370
- "uaddlt z7.h, z2.b, z3.b \n" \
371
- "add z4.h, z4.h, z6.h \n" \
372
- "add z5.h, z5.h, z7.h \n" \
373
- "rshrnb z0.b, z4.h, #2 \n" \
374
- "rshrnt z0.b, z5.h, #2 \n" \
375
- "st1h {z0.h}, p0, [%[dst_uv]] \n" \
376
- "incb %[dst_uv], all, mul #1 \n"
377
-
378
- __arm_locally_streaming void ScaleUVRowDown2Box_SME(const uint8_t* src_uv,
379
- ptrdiff_t src_stride,
380
- uint8_t* dst_uv,
381
- int dst_width) {
382
- // Streaming-SVE only, no use of ZA tile.
383
- const uint8_t* src2_uv = src_uv + src_stride;
384
- int vl;
385
- asm volatile(
386
- "cnth %x[vl] \n"
387
- "ptrue p1.b \n"
388
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
389
- "b.lt 2f \n"
390
-
391
- "ptrue p0.h \n"
392
- "1: \n" //
393
- SCALEUVROWDOWN2BOX_SVE
394
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
395
- "b.ge 1b \n"
396
-
397
- "2: \n"
398
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
399
- "b.eq 99f \n"
400
-
401
- "whilelt p0.h, wzr, %w[dst_width] \n" //
402
- SCALEUVROWDOWN2BOX_SVE
403
-
404
- "99: \n"
405
- : [src_uv] "+r"(src_uv), // %[src_uv]
406
- [src2_uv] "+r"(src2_uv), // %[src2_uv]
407
- [dst_uv] "+r"(dst_uv), // %[dst_uv]
408
- [dst_width] "+r"(dst_width), // %[dst_width]
409
- [vl] "=r"(vl) // %[vl]
410
- :
411
- : "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0", "p1");
412
- }
413
-
414
- #undef SCALEUVROWDOWN2BOX_SVE
415
-
416
- __arm_locally_streaming void ScaleARGBRowDown2_SME(const uint8_t* src_argb,
417
- ptrdiff_t src_stride,
418
- uint8_t* dst_argb,
419
- int dst_width) {
420
- // Streaming-SVE only, no use of ZA tile.
421
- (void)src_stride;
422
- int vl;
423
- asm volatile(
424
- "cntw %x[vl] \n"
425
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
426
- "b.lt 2f \n"
427
-
428
- "1: \n"
429
- "ptrue p0.b \n"
430
- "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
431
- "incb %[src_argb], all, mul #2 \n"
432
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
433
- "st1w {z1.s}, p0, [%[dst_argb]] \n"
434
- "incb %[dst_argb] \n"
435
- "b.ge 1b \n"
436
-
437
- "2: \n"
438
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
439
- "b.eq 99f \n"
440
-
441
- "whilelt p0.s, wzr, %w[dst_width] \n"
442
- "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
443
- "st1w {z1.s}, p0, [%[dst_argb]] \n"
444
-
445
- "99: \n"
446
- : [src_argb] "+r"(src_argb), // %[src_argb]
447
- [dst_argb] "+r"(dst_argb), // %[dst_argb]
448
- [dst_width] "+r"(dst_width), // %[dst_width]
449
- [vl] "=r"(vl) // %[vl]
450
- :
451
- : "memory", "cc", "z0", "z1", "p0");
452
- }
453
-
454
- __arm_locally_streaming void ScaleARGBRowDown2Linear_SME(
455
- const uint8_t* src_argb,
456
- ptrdiff_t src_stride,
457
- uint8_t* dst_argb,
458
- int dst_width) {
459
- // Streaming-SVE only, no use of ZA tile.
460
- (void)src_stride;
461
- int vl;
462
- asm volatile(
463
- "cntw %x[vl] \n"
464
- "ptrue p1.b \n"
465
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
466
- "b.lt 2f \n"
467
-
468
- "ptrue p0.s \n"
469
- "1: \n"
470
- "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
471
- "incb %[src_argb], all, mul #2 \n"
472
- "urhadd z0.b, p1/m, z0.b, z1.b \n"
473
- "st1w {z0.s}, p0, [%[dst_argb]] \n"
474
- "incb %[dst_argb], all, mul #1 \n"
475
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
476
- "b.ge 1b \n"
477
-
478
- "2: \n"
479
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
480
- "b.eq 99f \n"
481
-
482
- "whilelt p0.s, wzr, %w[dst_width] \n"
483
- "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
484
- "urhadd z0.b, p1/m, z0.b, z1.b \n"
485
- "st1w {z0.s}, p0, [%[dst_argb]] \n"
486
-
487
- "99: \n"
488
- : [src_argb] "+r"(src_argb), // %[src_argb]
489
- [dst_argb] "+r"(dst_argb), // %[dst_argb]
490
- [dst_width] "+r"(dst_width), // %[dst_width]
491
- [vl] "=r"(vl) // %[vl]
492
- :
493
- : "memory", "cc", "z0", "z1", "p0", "p1");
494
- }
495
-
496
- #define SCALEARGBROWDOWN2BOX_SVE \
497
- "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n" \
498
- "ld2w {z2.s, z3.s}, p0/z, [%[src2_argb]] \n" \
499
- "incb %[src_argb], all, mul #2 \n" \
500
- "incb %[src2_argb], all, mul #2 \n" \
501
- "uaddlb z4.h, z0.b, z1.b \n" \
502
- "uaddlt z5.h, z0.b, z1.b \n" \
503
- "uaddlb z6.h, z2.b, z3.b \n" \
504
- "uaddlt z7.h, z2.b, z3.b \n" \
505
- "add z4.h, z4.h, z6.h \n" \
506
- "add z5.h, z5.h, z7.h \n" \
507
- "rshrnb z0.b, z4.h, #2 \n" \
508
- "rshrnt z0.b, z5.h, #2 \n" \
509
- "st1w {z0.s}, p0, [%[dst_argb]] \n" \
510
- "incb %[dst_argb], all, mul #1 \n"
511
-
512
- __arm_locally_streaming void ScaleARGBRowDown2Box_SME(const uint8_t* src_argb,
513
- ptrdiff_t src_stride,
514
- uint8_t* dst_argb,
515
- int dst_width) {
516
- // Streaming-SVE only, no use of ZA tile.
517
- const uint8_t* src2_argb = src_argb + src_stride;
518
- int vl;
519
- asm volatile(
520
- "cntw %x[vl] \n"
521
- "ptrue p1.b \n"
522
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
523
- "b.lt 2f \n"
524
-
525
- "ptrue p0.s \n"
526
- "1: \n" //
527
- SCALEARGBROWDOWN2BOX_SVE
528
- "subs %w[dst_width], %w[dst_width], %w[vl] \n"
529
- "b.ge 1b \n"
530
-
531
- "2: \n"
532
- "adds %w[dst_width], %w[dst_width], %w[vl] \n"
533
- "b.eq 99f \n"
534
-
535
- "whilelt p0.s, wzr, %w[dst_width] \n" //
536
- SCALEARGBROWDOWN2BOX_SVE
537
-
538
- "99: \n"
539
- : [src_argb] "+r"(src_argb), // %[src_argb]
540
- [src2_argb] "+r"(src2_argb), // %[src2_argb]
541
- [dst_argb] "+r"(dst_argb), // %[dst_argb]
542
- [dst_width] "+r"(dst_width), // %[dst_width]
543
- [vl] "=r"(vl) // %[vl]
544
- :
545
- : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0",
546
- "p1");
547
- }
548
-
549
- #endif // !defined(LIBYUV_DISABLE_SME) && defined(CLANG_HAS_SME) &&
550
- // defined(__aarch64__)
551
-
552
- #ifdef __cplusplus
553
- } // extern "C"
554
- } // namespace libyuv
555
- #endif
1
+ /*
2
+ * Copyright 2024 The LibYuv Project Authors. All rights reserved.
3
+ *
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
6
+ * tree. An additional intellectual property rights grant can be found
7
+ * in the file PATENTS. All contributing project authors may
8
+ * be found in the AUTHORS file in the root of the source tree.
9
+ */
10
+
11
+ #include "libyuv/scale_row.h"
12
+
13
+ #ifdef __cplusplus
14
+ namespace libyuv {
15
+ extern "C" {
16
+ #endif
17
+
18
+ #if !defined(LIBYUV_DISABLE_SME) && defined(CLANG_HAS_SME) && \
19
+ defined(__aarch64__)
20
+
21
+ __arm_locally_streaming void ScaleRowDown2_SME(const uint8_t* src_ptr,
22
+ ptrdiff_t src_stride,
23
+ uint8_t* dst,
24
+ int dst_width) {
25
+ // Streaming-SVE only, no use of ZA tile.
26
+ (void)src_stride;
27
+ int vl;
28
+ asm volatile(
29
+ "cntb %x[vl] \n"
30
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
31
+ "b.lt 2f \n"
32
+
33
+ "1: \n"
34
+ "ptrue p0.b \n"
35
+ "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
36
+ "incb %[src_ptr], all, mul #2 \n"
37
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
38
+ "st1b {z1.b}, p0, [%[dst_ptr]] \n"
39
+ "incb %[dst_ptr] \n"
40
+ "b.ge 1b \n"
41
+
42
+ "2: \n"
43
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
44
+ "b.eq 99f \n"
45
+
46
+ "whilelt p0.b, wzr, %w[dst_width] \n"
47
+ "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
48
+ "st1b {z1.b}, p0, [%[dst_ptr]] \n"
49
+
50
+ "99: \n"
51
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
52
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
53
+ [dst_width] "+r"(dst_width), // %[dst_width]
54
+ [vl] "=r"(vl) // %[vl]
55
+ :
56
+ : "memory", "cc", "z0", "z1", "p0");
57
+ }
58
+
59
+ __arm_locally_streaming void ScaleRowDown2_16_SME(const uint16_t* src_ptr,
60
+ ptrdiff_t src_stride,
61
+ uint16_t* dst,
62
+ int dst_width) {
63
+ // Streaming-SVE only, no use of ZA tile.
64
+ (void)src_stride;
65
+ int vl;
66
+ asm volatile(
67
+ "cnth %x[vl] \n"
68
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
69
+ "b.lt 2f \n"
70
+
71
+ "1: \n"
72
+ "ptrue p0.h \n"
73
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
74
+ "incb %[src_ptr], all, mul #2 \n"
75
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
76
+ "st1h {z1.h}, p0, [%[dst_ptr]] \n"
77
+ "incb %[dst_ptr] \n"
78
+ "b.ge 1b \n"
79
+
80
+ "2: \n"
81
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
82
+ "b.eq 99f \n"
83
+
84
+ "whilelt p0.h, wzr, %w[dst_width] \n"
85
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
86
+ "st1h {z1.h}, p0, [%[dst_ptr]] \n"
87
+
88
+ "99: \n"
89
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
90
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
91
+ [dst_width] "+r"(dst_width), // %[dst_width]
92
+ [vl] "=r"(vl) // %[vl]
93
+ :
94
+ : "memory", "cc", "z0", "z1", "p0");
95
+ }
96
+
97
+ __arm_locally_streaming void ScaleRowDown2Linear_SME(const uint8_t* src_ptr,
98
+ ptrdiff_t src_stride,
99
+ uint8_t* dst,
100
+ int dst_width) {
101
+ // Streaming-SVE only, no use of ZA tile.
102
+ (void)src_stride;
103
+ int vl;
104
+ asm volatile(
105
+ "cntb %x[vl] \n"
106
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
107
+ "b.lt 2f \n"
108
+
109
+ "1: \n"
110
+ "ptrue p0.b \n"
111
+ "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
112
+ "incb %[src_ptr], all, mul #2 \n"
113
+ "urhadd z0.b, p0/m, z0.b, z1.b \n"
114
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
115
+ "st1b {z0.b}, p0, [%[dst_ptr]] \n"
116
+ "incb %[dst_ptr] \n"
117
+ "b.ge 1b \n"
118
+
119
+ "2: \n"
120
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
121
+ "b.eq 99f \n"
122
+
123
+ "whilelt p0.b, wzr, %w[dst_width] \n"
124
+ "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n"
125
+ "urhadd z0.b, p0/m, z0.b, z1.b \n"
126
+ "st1b {z0.b}, p0, [%[dst_ptr]] \n"
127
+
128
+ "99: \n"
129
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
130
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
131
+ [dst_width] "+r"(dst_width), // %[dst_width]
132
+ [vl] "=r"(vl) // %[vl]
133
+ :
134
+ : "memory", "cc", "z0", "z1", "p0");
135
+ }
136
+
137
+ __arm_locally_streaming void ScaleRowDown2Linear_16_SME(const uint16_t* src_ptr,
138
+ ptrdiff_t src_stride,
139
+ uint16_t* dst,
140
+ int dst_width) {
141
+ // Streaming-SVE only, no use of ZA tile.
142
+ (void)src_stride;
143
+ int vl;
144
+ asm volatile(
145
+ "cnth %x[vl] \n"
146
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
147
+ "b.lt 2f \n"
148
+
149
+ "1: \n"
150
+ "ptrue p0.h \n"
151
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
152
+ "incb %[src_ptr], all, mul #2 \n"
153
+ "urhadd z0.h, p0/m, z0.h, z1.h \n"
154
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
155
+ "st1h {z0.h}, p0, [%[dst_ptr]] \n"
156
+ "incb %[dst_ptr] \n"
157
+ "b.ge 1b \n"
158
+
159
+ "2: \n"
160
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
161
+ "b.eq 99f \n"
162
+
163
+ "whilelt p0.h, wzr, %w[dst_width] \n"
164
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n"
165
+ "urhadd z0.h, p0/m, z0.h, z1.h \n"
166
+ "st1h {z0.h}, p0, [%[dst_ptr]] \n"
167
+
168
+ "99: \n"
169
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
170
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
171
+ [dst_width] "+r"(dst_width), // %[dst_width]
172
+ [vl] "=r"(vl) // %[vl]
173
+ :
174
+ : "memory", "cc", "z0", "z1", "p0");
175
+ }
176
+
177
+ #define SCALEROWDOWN2BOX_SVE \
178
+ "ld2b {z0.b, z1.b}, p0/z, [%[src_ptr]] \n" \
179
+ "ld2b {z2.b, z3.b}, p0/z, [%[src2_ptr]] \n" \
180
+ "incb %[src_ptr], all, mul #2 \n" \
181
+ "incb %[src2_ptr], all, mul #2 \n" \
182
+ "uaddlb z4.h, z0.b, z1.b \n" \
183
+ "uaddlt z5.h, z0.b, z1.b \n" \
184
+ "uaddlb z6.h, z2.b, z3.b \n" \
185
+ "uaddlt z7.h, z2.b, z3.b \n" \
186
+ "add z4.h, z4.h, z6.h \n" \
187
+ "add z5.h, z5.h, z7.h \n" \
188
+ "rshrnb z0.b, z4.h, #2 \n" \
189
+ "rshrnt z0.b, z5.h, #2 \n" \
190
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n" \
191
+ "st1b {z0.b}, p0, [%[dst_ptr]] \n" \
192
+ "incb %[dst_ptr] \n"
193
+
194
+ __arm_locally_streaming void ScaleRowDown2Box_SME(const uint8_t* src_ptr,
195
+ ptrdiff_t src_stride,
196
+ uint8_t* dst,
197
+ int dst_width) {
198
+ // Streaming-SVE only, no use of ZA tile.
199
+ const uint8_t* src2_ptr = src_ptr + src_stride;
200
+ int vl;
201
+ asm volatile(
202
+ "cntb %x[vl] \n"
203
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
204
+ "b.lt 2f \n"
205
+
206
+ "ptrue p0.b \n"
207
+ "1: \n" //
208
+ SCALEROWDOWN2BOX_SVE
209
+ "b.ge 1b \n"
210
+
211
+ "2: \n"
212
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
213
+ "b.eq 99f \n"
214
+
215
+ "whilelt p0.b, wzr, %w[dst_width] \n" //
216
+ SCALEROWDOWN2BOX_SVE
217
+
218
+ "99: \n"
219
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
220
+ [src2_ptr] "+r"(src2_ptr), // %[src2_ptr]
221
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
222
+ [dst_width] "+r"(dst_width), // %[dst_width]
223
+ [vl] "=r"(vl) // %[vl]
224
+ :
225
+ : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0");
226
+ }
227
+
228
+ #undef SCALEROWDOWN2BOX_SVE
229
+
230
+ #define SCALEROWDOWN2BOX_16_SVE \
231
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_ptr]] \n" \
232
+ "ld2h {z2.h, z3.h}, p0/z, [%[src2_ptr]] \n" \
233
+ "incb %[src_ptr], all, mul #2 \n" \
234
+ "incb %[src2_ptr], all, mul #2 \n" \
235
+ "uaddlb z4.s, z0.h, z1.h \n" \
236
+ "uaddlt z5.s, z0.h, z1.h \n" \
237
+ "uaddlb z6.s, z2.h, z3.h \n" \
238
+ "uaddlt z7.s, z2.h, z3.h \n" \
239
+ "add z4.s, z4.s, z6.s \n" \
240
+ "add z5.s, z5.s, z7.s \n" \
241
+ "rshrnb z0.h, z4.s, #2 \n" \
242
+ "rshrnt z0.h, z5.s, #2 \n" \
243
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n" \
244
+ "st1h {z0.h}, p0, [%[dst_ptr]] \n" \
245
+ "incb %[dst_ptr] \n"
246
+
247
+ __arm_locally_streaming void ScaleRowDown2Box_16_SME(const uint16_t* src_ptr,
248
+ ptrdiff_t src_stride,
249
+ uint16_t* dst,
250
+ int dst_width) {
251
+ // Streaming-SVE only, no use of ZA tile.
252
+ const uint16_t* src2_ptr = src_ptr + src_stride;
253
+ int vl;
254
+ asm volatile(
255
+ "cnth %x[vl] \n"
256
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
257
+ "b.lt 2f \n"
258
+
259
+ "ptrue p0.h \n"
260
+ "1: \n" //
261
+ SCALEROWDOWN2BOX_16_SVE
262
+ "b.ge 1b \n"
263
+
264
+ "2: \n"
265
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
266
+ "b.eq 99f \n"
267
+
268
+ "whilelt p0.h, wzr, %w[dst_width] \n" //
269
+ SCALEROWDOWN2BOX_16_SVE
270
+
271
+ "99: \n"
272
+ : [src_ptr] "+r"(src_ptr), // %[src_ptr]
273
+ [src2_ptr] "+r"(src2_ptr), // %[src2_ptr]
274
+ [dst_ptr] "+r"(dst), // %[dst_ptr]
275
+ [dst_width] "+r"(dst_width), // %[dst_width]
276
+ [vl] "=r"(vl) // %[vl]
277
+ :
278
+ : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0");
279
+ }
280
+
281
+ #undef SCALEROWDOWN2BOX_16_SVE
282
+
283
+ __arm_locally_streaming void ScaleUVRowDown2_SME(const uint8_t* src_uv,
284
+ ptrdiff_t src_stride,
285
+ uint8_t* dst_uv,
286
+ int dst_width) {
287
+ // Streaming-SVE only, no use of ZA tile.
288
+ (void)src_stride;
289
+ int vl;
290
+ asm volatile(
291
+ "cnth %x[vl] \n"
292
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
293
+ "b.lt 2f \n"
294
+
295
+ "1: \n"
296
+ "ptrue p0.b \n"
297
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
298
+ "incb %[src_uv], all, mul #2 \n"
299
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
300
+ "st1h {z1.h}, p0, [%[dst_uv]] \n"
301
+ "incb %[dst_uv] \n"
302
+ "b.ge 1b \n"
303
+
304
+ "2: \n"
305
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
306
+ "b.eq 99f \n"
307
+
308
+ "whilelt p0.h, wzr, %w[dst_width] \n"
309
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
310
+ "st1h {z1.h}, p0, [%[dst_uv]] \n"
311
+
312
+ "99: \n"
313
+ : [src_uv] "+r"(src_uv), // %[src_uv]
314
+ [dst_uv] "+r"(dst_uv), // %[dst_uv]
315
+ [dst_width] "+r"(dst_width), // %[dst_width]
316
+ [vl] "=r"(vl) // %[vl]
317
+ :
318
+ : "memory", "cc", "z0", "z1", "p0");
319
+ }
320
+
321
+ __arm_locally_streaming void ScaleUVRowDown2Linear_SME(const uint8_t* src_uv,
322
+ ptrdiff_t src_stride,
323
+ uint8_t* dst_uv,
324
+ int dst_width) {
325
+ // Streaming-SVE only, no use of ZA tile.
326
+ (void)src_stride;
327
+ int vl;
328
+ asm volatile(
329
+ "cnth %x[vl] \n"
330
+ "ptrue p1.b \n"
331
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
332
+ "b.lt 2f \n"
333
+
334
+ "ptrue p0.h \n"
335
+ "1: \n"
336
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
337
+ "incb %[src_uv], all, mul #2 \n"
338
+ "urhadd z0.b, p1/m, z0.b, z1.b \n"
339
+ "st1h {z0.h}, p0, [%[dst_uv]] \n"
340
+ "incb %[dst_uv], all, mul #1 \n"
341
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
342
+ "b.ge 1b \n"
343
+
344
+ "2: \n"
345
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
346
+ "b.eq 99f \n"
347
+
348
+ "whilelt p0.h, wzr, %w[dst_width] \n"
349
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n"
350
+ "urhadd z0.b, p1/m, z0.b, z1.b \n"
351
+ "st1h {z0.h}, p0, [%[dst_uv]] \n"
352
+
353
+ "99: \n"
354
+ : [src_uv] "+r"(src_uv), // %[src_uv]
355
+ [dst_uv] "+r"(dst_uv), // %[dst_uv]
356
+ [dst_width] "+r"(dst_width), // %[dst_width]
357
+ [vl] "=r"(vl) // %[vl]
358
+ :
359
+ : "z0", "z1", "p0", "p1");
360
+ }
361
+
362
+ #define SCALEUVROWDOWN2BOX_SVE \
363
+ "ld2h {z0.h, z1.h}, p0/z, [%[src_uv]] \n" \
364
+ "ld2h {z2.h, z3.h}, p0/z, [%[src2_uv]] \n" \
365
+ "incb %[src_uv], all, mul #2 \n" \
366
+ "incb %[src2_uv], all, mul #2 \n" \
367
+ "uaddlb z4.h, z0.b, z1.b \n" \
368
+ "uaddlt z5.h, z0.b, z1.b \n" \
369
+ "uaddlb z6.h, z2.b, z3.b \n" \
370
+ "uaddlt z7.h, z2.b, z3.b \n" \
371
+ "add z4.h, z4.h, z6.h \n" \
372
+ "add z5.h, z5.h, z7.h \n" \
373
+ "rshrnb z0.b, z4.h, #2 \n" \
374
+ "rshrnt z0.b, z5.h, #2 \n" \
375
+ "st1h {z0.h}, p0, [%[dst_uv]] \n" \
376
+ "incb %[dst_uv], all, mul #1 \n"
377
+
378
+ __arm_locally_streaming void ScaleUVRowDown2Box_SME(const uint8_t* src_uv,
379
+ ptrdiff_t src_stride,
380
+ uint8_t* dst_uv,
381
+ int dst_width) {
382
+ // Streaming-SVE only, no use of ZA tile.
383
+ const uint8_t* src2_uv = src_uv + src_stride;
384
+ int vl;
385
+ asm volatile(
386
+ "cnth %x[vl] \n"
387
+ "ptrue p1.b \n"
388
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
389
+ "b.lt 2f \n"
390
+
391
+ "ptrue p0.h \n"
392
+ "1: \n" //
393
+ SCALEUVROWDOWN2BOX_SVE
394
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
395
+ "b.ge 1b \n"
396
+
397
+ "2: \n"
398
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
399
+ "b.eq 99f \n"
400
+
401
+ "whilelt p0.h, wzr, %w[dst_width] \n" //
402
+ SCALEUVROWDOWN2BOX_SVE
403
+
404
+ "99: \n"
405
+ : [src_uv] "+r"(src_uv), // %[src_uv]
406
+ [src2_uv] "+r"(src2_uv), // %[src2_uv]
407
+ [dst_uv] "+r"(dst_uv), // %[dst_uv]
408
+ [dst_width] "+r"(dst_width), // %[dst_width]
409
+ [vl] "=r"(vl) // %[vl]
410
+ :
411
+ : "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0", "p1");
412
+ }
413
+
414
+ #undef SCALEUVROWDOWN2BOX_SVE
415
+
416
+ __arm_locally_streaming void ScaleARGBRowDown2_SME(const uint8_t* src_argb,
417
+ ptrdiff_t src_stride,
418
+ uint8_t* dst_argb,
419
+ int dst_width) {
420
+ // Streaming-SVE only, no use of ZA tile.
421
+ (void)src_stride;
422
+ int vl;
423
+ asm volatile(
424
+ "cntw %x[vl] \n"
425
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
426
+ "b.lt 2f \n"
427
+
428
+ "1: \n"
429
+ "ptrue p0.b \n"
430
+ "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
431
+ "incb %[src_argb], all, mul #2 \n"
432
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
433
+ "st1w {z1.s}, p0, [%[dst_argb]] \n"
434
+ "incb %[dst_argb] \n"
435
+ "b.ge 1b \n"
436
+
437
+ "2: \n"
438
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
439
+ "b.eq 99f \n"
440
+
441
+ "whilelt p0.s, wzr, %w[dst_width] \n"
442
+ "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
443
+ "st1w {z1.s}, p0, [%[dst_argb]] \n"
444
+
445
+ "99: \n"
446
+ : [src_argb] "+r"(src_argb), // %[src_argb]
447
+ [dst_argb] "+r"(dst_argb), // %[dst_argb]
448
+ [dst_width] "+r"(dst_width), // %[dst_width]
449
+ [vl] "=r"(vl) // %[vl]
450
+ :
451
+ : "memory", "cc", "z0", "z1", "p0");
452
+ }
453
+
454
+ __arm_locally_streaming void ScaleARGBRowDown2Linear_SME(
455
+ const uint8_t* src_argb,
456
+ ptrdiff_t src_stride,
457
+ uint8_t* dst_argb,
458
+ int dst_width) {
459
+ // Streaming-SVE only, no use of ZA tile.
460
+ (void)src_stride;
461
+ int vl;
462
+ asm volatile(
463
+ "cntw %x[vl] \n"
464
+ "ptrue p1.b \n"
465
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
466
+ "b.lt 2f \n"
467
+
468
+ "ptrue p0.s \n"
469
+ "1: \n"
470
+ "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
471
+ "incb %[src_argb], all, mul #2 \n"
472
+ "urhadd z0.b, p1/m, z0.b, z1.b \n"
473
+ "st1w {z0.s}, p0, [%[dst_argb]] \n"
474
+ "incb %[dst_argb], all, mul #1 \n"
475
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
476
+ "b.ge 1b \n"
477
+
478
+ "2: \n"
479
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
480
+ "b.eq 99f \n"
481
+
482
+ "whilelt p0.s, wzr, %w[dst_width] \n"
483
+ "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n"
484
+ "urhadd z0.b, p1/m, z0.b, z1.b \n"
485
+ "st1w {z0.s}, p0, [%[dst_argb]] \n"
486
+
487
+ "99: \n"
488
+ : [src_argb] "+r"(src_argb), // %[src_argb]
489
+ [dst_argb] "+r"(dst_argb), // %[dst_argb]
490
+ [dst_width] "+r"(dst_width), // %[dst_width]
491
+ [vl] "=r"(vl) // %[vl]
492
+ :
493
+ : "memory", "cc", "z0", "z1", "p0", "p1");
494
+ }
495
+
496
+ #define SCALEARGBROWDOWN2BOX_SVE \
497
+ "ld2w {z0.s, z1.s}, p0/z, [%[src_argb]] \n" \
498
+ "ld2w {z2.s, z3.s}, p0/z, [%[src2_argb]] \n" \
499
+ "incb %[src_argb], all, mul #2 \n" \
500
+ "incb %[src2_argb], all, mul #2 \n" \
501
+ "uaddlb z4.h, z0.b, z1.b \n" \
502
+ "uaddlt z5.h, z0.b, z1.b \n" \
503
+ "uaddlb z6.h, z2.b, z3.b \n" \
504
+ "uaddlt z7.h, z2.b, z3.b \n" \
505
+ "add z4.h, z4.h, z6.h \n" \
506
+ "add z5.h, z5.h, z7.h \n" \
507
+ "rshrnb z0.b, z4.h, #2 \n" \
508
+ "rshrnt z0.b, z5.h, #2 \n" \
509
+ "st1w {z0.s}, p0, [%[dst_argb]] \n" \
510
+ "incb %[dst_argb], all, mul #1 \n"
511
+
512
+ __arm_locally_streaming void ScaleARGBRowDown2Box_SME(const uint8_t* src_argb,
513
+ ptrdiff_t src_stride,
514
+ uint8_t* dst_argb,
515
+ int dst_width) {
516
+ // Streaming-SVE only, no use of ZA tile.
517
+ const uint8_t* src2_argb = src_argb + src_stride;
518
+ int vl;
519
+ asm volatile(
520
+ "cntw %x[vl] \n"
521
+ "ptrue p1.b \n"
522
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
523
+ "b.lt 2f \n"
524
+
525
+ "ptrue p0.s \n"
526
+ "1: \n" //
527
+ SCALEARGBROWDOWN2BOX_SVE
528
+ "subs %w[dst_width], %w[dst_width], %w[vl] \n"
529
+ "b.ge 1b \n"
530
+
531
+ "2: \n"
532
+ "adds %w[dst_width], %w[dst_width], %w[vl] \n"
533
+ "b.eq 99f \n"
534
+
535
+ "whilelt p0.s, wzr, %w[dst_width] \n" //
536
+ SCALEARGBROWDOWN2BOX_SVE
537
+
538
+ "99: \n"
539
+ : [src_argb] "+r"(src_argb), // %[src_argb]
540
+ [src2_argb] "+r"(src2_argb), // %[src2_argb]
541
+ [dst_argb] "+r"(dst_argb), // %[dst_argb]
542
+ [dst_width] "+r"(dst_width), // %[dst_width]
543
+ [vl] "=r"(vl) // %[vl]
544
+ :
545
+ : "memory", "cc", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "p0",
546
+ "p1");
547
+ }
548
+
549
+ #endif // !defined(LIBYUV_DISABLE_SME) && defined(CLANG_HAS_SME) &&
550
+ // defined(__aarch64__)
551
+
552
+ #ifdef __cplusplus
553
+ } // extern "C"
554
+ } // namespace libyuv
555
+ #endif