| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * This file is part of FFmpeg. | ||
| 3 | * | ||
| 4 | * FFmpeg is free software; you can redistribute it and/or | ||
| 5 | * modify it under the terms of the GNU Lesser General Public | ||
| 6 | * License as published by the Free Software Foundation; either | ||
| 7 | * version 2.1 of the License, or (at your option) any later version. | ||
| 8 | * | ||
| 9 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
| 12 | * Lesser General Public License for more details. | ||
| 13 | * | ||
| 14 | * You should have received a copy of the GNU Lesser General Public | ||
| 15 | * License along with FFmpeg; if not, write to the Free Software | ||
| 16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | ||
| 17 | */ | ||
| 18 | |||
| 19 | #include <assert.h> | ||
| 20 | #include <stdint.h> | ||
| 21 | #include <string.h> | ||
| 22 | |||
| 23 | #include "config.h" | ||
| 24 | #include "libavutil/avassert.h" | ||
| 25 | #include "libavutil/attributes.h" | ||
| 26 | #include "libavutil/imgutils.h" | ||
| 27 | #include "avcodec.h" | ||
| 28 | #include "me_cmp.h" | ||
| 29 | #include "mpegvideoencdsp.h" | ||
| 30 | |||
| 31 | ✗ | static int try_8x8basis_c(const int16_t rem[64], const int16_t weight[64], | |
| 32 | const int16_t basis[64], int scale) | ||
| 33 | { | ||
| 34 | int i; | ||
| 35 | ✗ | unsigned int sum = 0; | |
| 36 | |||
| 37 | ✗ | for (i = 0; i < 8 * 8; i++) { | |
| 38 | ✗ | int b = rem[i] + ((basis[i] * scale + | |
| 39 | ✗ | (1 << (BASIS_SHIFT - RECON_SHIFT - 1))) >> | |
| 40 | (BASIS_SHIFT - RECON_SHIFT)); | ||
| 41 | ✗ | int w = weight[i]; | |
| 42 | ✗ | b >>= RECON_SHIFT; | |
| 43 | av_assert2(-512 < b && b < 512); | ||
| 44 | |||
| 45 | ✗ | sum += (w * b) * (w * b) >> 4; | |
| 46 | } | ||
| 47 | ✗ | return sum >> 2; | |
| 48 | } | ||
| 49 | |||
| 50 | ✗ | static void add_8x8basis_c(int16_t rem[64], const int16_t basis[64], int scale) | |
| 51 | { | ||
| 52 | int i; | ||
| 53 | |||
| 54 | ✗ | for (i = 0; i < 8 * 8; i++) | |
| 55 | ✗ | rem[i] += (basis[i] * scale + | |
| 56 | ✗ | (1 << (BASIS_SHIFT - RECON_SHIFT - 1))) >> | |
| 57 | (BASIS_SHIFT - RECON_SHIFT); | ||
| 58 | ✗ | } | |
| 59 | |||
| 60 | 4341418 | static int pix_sum_c(const uint8_t *pix, ptrdiff_t line_size) | |
| 61 | { | ||
| 62 | 4341418 | int s = 0, i, j; | |
| 63 | |||
| 64 |
2/2✓ Branch 0 taken 69462688 times.
✓ Branch 1 taken 4341418 times.
|
73804106 | for (i = 0; i < 16; i++) { |
| 65 |
2/2✓ Branch 0 taken 138925376 times.
✓ Branch 1 taken 69462688 times.
|
208388064 | for (j = 0; j < 16; j += 8) { |
| 66 | 138925376 | s += pix[0]; | |
| 67 | 138925376 | s += pix[1]; | |
| 68 | 138925376 | s += pix[2]; | |
| 69 | 138925376 | s += pix[3]; | |
| 70 | 138925376 | s += pix[4]; | |
| 71 | 138925376 | s += pix[5]; | |
| 72 | 138925376 | s += pix[6]; | |
| 73 | 138925376 | s += pix[7]; | |
| 74 | 138925376 | pix += 8; | |
| 75 | } | ||
| 76 | 69462688 | pix += line_size - 16; | |
| 77 | } | ||
| 78 | 4341418 | return s; | |
| 79 | } | ||
| 80 | |||
| 81 | 4341418 | static int pix_norm1_c(const uint8_t *pix, ptrdiff_t line_size) | |
| 82 | { | ||
| 83 | 4341418 | int s = 0, i, j; | |
| 84 | 4341418 | const uint32_t *sq = ff_square_tab + 256; | |
| 85 | |||
| 86 |
2/2✓ Branch 0 taken 69462688 times.
✓ Branch 1 taken 4341418 times.
|
73804106 | for (i = 0; i < 16; i++) { |
| 87 |
2/2✓ Branch 0 taken 138925376 times.
✓ Branch 1 taken 69462688 times.
|
208388064 | for (j = 0; j < 16; j += 8) { |
| 88 | #if HAVE_FAST_64BIT | ||
| 89 | 138925376 | register uint64_t x = *(uint64_t *) pix; | |
| 90 | 138925376 | s += sq[x & 0xff]; | |
| 91 | 138925376 | s += sq[(x >> 8) & 0xff]; | |
| 92 | 138925376 | s += sq[(x >> 16) & 0xff]; | |
| 93 | 138925376 | s += sq[(x >> 24) & 0xff]; | |
| 94 | 138925376 | s += sq[(x >> 32) & 0xff]; | |
| 95 | 138925376 | s += sq[(x >> 40) & 0xff]; | |
| 96 | 138925376 | s += sq[(x >> 48) & 0xff]; | |
| 97 | 138925376 | s += sq[(x >> 56) & 0xff]; | |
| 98 | #else | ||
| 99 | register uint32_t x = *(uint32_t *) pix; | ||
| 100 | s += sq[x & 0xff]; | ||
| 101 | s += sq[(x >> 8) & 0xff]; | ||
| 102 | s += sq[(x >> 16) & 0xff]; | ||
| 103 | s += sq[(x >> 24) & 0xff]; | ||
| 104 | x = *(uint32_t *) (pix + 4); | ||
| 105 | s += sq[x & 0xff]; | ||
| 106 | s += sq[(x >> 8) & 0xff]; | ||
| 107 | s += sq[(x >> 16) & 0xff]; | ||
| 108 | s += sq[(x >> 24) & 0xff]; | ||
| 109 | #endif | ||
| 110 | 138925376 | pix += 8; | |
| 111 | } | ||
| 112 | 69462688 | pix += line_size - 16; | |
| 113 | } | ||
| 114 | 4341418 | return s; | |
| 115 | } | ||
| 116 | |||
| 117 | 21201 | static av_always_inline void draw_edges_lr(uint8_t *ptr, ptrdiff_t wrap, int width, int height, int w) | |
| 118 | { | ||
| 119 |
2/2✓ Branch 0 taken 2199914 times.
✓ Branch 1 taken 21201 times.
|
2221115 | for (int i = 0; i < height; i++) { |
| 120 | 2199914 | memset(ptr - w, ptr[0], w); | |
| 121 | 2199914 | memset(ptr + width, ptr[width - 1], w); | |
| 122 | 2199914 | ptr += wrap; | |
| 123 | } | ||
| 124 | 21201 | } | |
| 125 | |||
| 126 | /* draw the edges of width 'w' of an image of size width, height */ | ||
| 127 | // FIXME: Check that this is OK for MPEG-4 interlaced. | ||
| 128 | 21201 | static void draw_edges_8_c(uint8_t *buf, ptrdiff_t wrap, int width, int height, | |
| 129 | int w, int h, int sides) | ||
| 130 | { | ||
| 131 | uint8_t *last_line; | ||
| 132 | int i; | ||
| 133 | |||
| 134 | /* left and right */ | ||
| 135 |
2/2✓ Branch 0 taken 7110 times.
✓ Branch 1 taken 14091 times.
|
21201 | if (w == 16) { |
| 136 | 7110 | draw_edges_lr(buf, wrap, width, height, 16); | |
| 137 |
2/2✓ Branch 0 taken 14076 times.
✓ Branch 1 taken 15 times.
|
14091 | } else if (w == 8) { |
| 138 | 14076 | draw_edges_lr(buf, wrap, width, height, 8); | |
| 139 | } else { | ||
| 140 | av_assert1(w == 4); | ||
| 141 | 15 | draw_edges_lr(buf, wrap, width, height, 4); | |
| 142 | } | ||
| 143 | |||
| 144 | /* top and bottom + corners */ | ||
| 145 | 21201 | buf -= w; | |
| 146 | 21201 | last_line = buf + (height - 1) * wrap; | |
| 147 |
2/2✓ Branch 0 taken 13263 times.
✓ Branch 1 taken 7938 times.
|
21201 | if (sides & EDGE_TOP) |
| 148 |
2/2✓ Branch 0 taken 140956 times.
✓ Branch 1 taken 13263 times.
|
154219 | for (i = 0; i < h; i++) |
| 149 | // top | ||
| 150 | 140956 | memcpy(buf - (i + 1) * wrap, buf, width + w + w); | |
| 151 |
1/2✓ Branch 0 taken 21201 times.
✗ Branch 1 not taken.
|
21201 | if (sides & EDGE_BOTTOM) |
| 152 |
2/2✓ Branch 0 taken 235228 times.
✓ Branch 1 taken 21201 times.
|
256429 | for (i = 0; i < h; i++) |
| 153 | // bottom | ||
| 154 | 235228 | memcpy(last_line + (i + 1) * wrap, last_line, width + w + w); | |
| 155 | 21201 | } | |
| 156 | |||
| 157 | /* This wrapper function only serves to convert the stride parameters | ||
| 158 | * from ptrdiff_t to int for av_image_copy_plane(). */ | ||
| 159 | ✗ | static void copy_plane_wrapper(uint8_t *dst, ptrdiff_t dst_wrap, | |
| 160 | const uint8_t *src, ptrdiff_t src_wrap, | ||
| 161 | int width, int height) | ||
| 162 | { | ||
| 163 | ✗ | av_image_copy_plane(dst, dst_wrap, src, src_wrap, width, height); | |
| 164 | ✗ | } | |
| 165 | |||
| 166 | /* 2x2 -> 1x1 */ | ||
| 167 | ✗ | static void shrink22(uint8_t *dst, ptrdiff_t dst_wrap, | |
| 168 | const uint8_t *src, ptrdiff_t src_wrap, | ||
| 169 | int width, int height) | ||
| 170 | { | ||
| 171 | int w; | ||
| 172 | const uint8_t *s1, *s2; | ||
| 173 | uint8_t *d; | ||
| 174 | |||
| 175 | ✗ | for (; height > 0; height--) { | |
| 176 | ✗ | s1 = src; | |
| 177 | ✗ | s2 = s1 + src_wrap; | |
| 178 | ✗ | d = dst; | |
| 179 | ✗ | for (w = width; w >= 4; w -= 4) { | |
| 180 | ✗ | d[0] = (s1[0] + s1[1] + s2[0] + s2[1] + 2) >> 2; | |
| 181 | ✗ | d[1] = (s1[2] + s1[3] + s2[2] + s2[3] + 2) >> 2; | |
| 182 | ✗ | d[2] = (s1[4] + s1[5] + s2[4] + s2[5] + 2) >> 2; | |
| 183 | ✗ | d[3] = (s1[6] + s1[7] + s2[6] + s2[7] + 2) >> 2; | |
| 184 | ✗ | s1 += 8; | |
| 185 | ✗ | s2 += 8; | |
| 186 | ✗ | d += 4; | |
| 187 | } | ||
| 188 | ✗ | for (; w > 0; w--) { | |
| 189 | ✗ | d[0] = (s1[0] + s1[1] + s2[0] + s2[1] + 2) >> 2; | |
| 190 | ✗ | s1 += 2; | |
| 191 | ✗ | s2 += 2; | |
| 192 | ✗ | d++; | |
| 193 | } | ||
| 194 | ✗ | src += 2 * src_wrap; | |
| 195 | ✗ | dst += dst_wrap; | |
| 196 | } | ||
| 197 | ✗ | } | |
| 198 | |||
| 199 | /* 4x4 -> 1x1 */ | ||
| 200 | ✗ | static void shrink44(uint8_t *dst, ptrdiff_t dst_wrap, | |
| 201 | const uint8_t *src, ptrdiff_t src_wrap, | ||
| 202 | int width, int height) | ||
| 203 | { | ||
| 204 | int w; | ||
| 205 | const uint8_t *s1, *s2, *s3, *s4; | ||
| 206 | uint8_t *d; | ||
| 207 | |||
| 208 | ✗ | for (; height > 0; height--) { | |
| 209 | ✗ | s1 = src; | |
| 210 | ✗ | s2 = s1 + src_wrap; | |
| 211 | ✗ | s3 = s2 + src_wrap; | |
| 212 | ✗ | s4 = s3 + src_wrap; | |
| 213 | ✗ | d = dst; | |
| 214 | ✗ | for (w = width; w > 0; w--) { | |
| 215 | ✗ | d[0] = (s1[0] + s1[1] + s1[2] + s1[3] + | |
| 216 | ✗ | s2[0] + s2[1] + s2[2] + s2[3] + | |
| 217 | ✗ | s3[0] + s3[1] + s3[2] + s3[3] + | |
| 218 | ✗ | s4[0] + s4[1] + s4[2] + s4[3] + 8) >> 4; | |
| 219 | ✗ | s1 += 4; | |
| 220 | ✗ | s2 += 4; | |
| 221 | ✗ | s3 += 4; | |
| 222 | ✗ | s4 += 4; | |
| 223 | ✗ | d++; | |
| 224 | } | ||
| 225 | ✗ | src += 4 * src_wrap; | |
| 226 | ✗ | dst += dst_wrap; | |
| 227 | } | ||
| 228 | ✗ | } | |
| 229 | |||
| 230 | /* 8x8 -> 1x1 */ | ||
| 231 | ✗ | static void shrink88(uint8_t *dst, ptrdiff_t dst_wrap, | |
| 232 | const uint8_t *src, ptrdiff_t src_wrap, | ||
| 233 | int width, int height) | ||
| 234 | { | ||
| 235 | int w, i; | ||
| 236 | |||
| 237 | ✗ | for (; height > 0; height--) { | |
| 238 | ✗ | for(w = width;w > 0; w--) { | |
| 239 | ✗ | int tmp = 0; | |
| 240 | ✗ | for (i = 0; i < 8; i++) { | |
| 241 | ✗ | tmp += src[0] + src[1] + src[2] + src[3] + | |
| 242 | ✗ | src[4] + src[5] + src[6] + src[7]; | |
| 243 | ✗ | src += src_wrap; | |
| 244 | } | ||
| 245 | ✗ | *(dst++) = (tmp + 32) >> 6; | |
| 246 | ✗ | src += 8 - 8 * src_wrap; | |
| 247 | } | ||
| 248 | ✗ | src += 8 * src_wrap - 8 * width; | |
| 249 | ✗ | dst += dst_wrap - width; | |
| 250 | } | ||
| 251 | ✗ | } | |
| 252 | |||
| 253 | 375 | av_cold void ff_mpegvideoencdsp_init(MpegvideoEncDSPContext *c, | |
| 254 | AVCodecContext *avctx) | ||
| 255 | { | ||
| 256 | 375 | c->try_8x8basis = try_8x8basis_c; | |
| 257 | 375 | c->add_8x8basis = add_8x8basis_c; | |
| 258 | |||
| 259 | 375 | c->shrink[0] = copy_plane_wrapper; | |
| 260 | 375 | c->shrink[1] = shrink22; | |
| 261 | 375 | c->shrink[2] = shrink44; | |
| 262 | 375 | c->shrink[3] = shrink88; | |
| 263 | |||
| 264 | 375 | c->pix_sum = pix_sum_c; | |
| 265 | 375 | c->pix_norm1 = pix_norm1_c; | |
| 266 | |||
| 267 | 375 | c->draw_edges = draw_edges_8_c; | |
| 268 | |||
| 269 | #if ARCH_AARCH64 | ||
| 270 | ff_mpegvideoencdsp_init_aarch64(c, avctx); | ||
| 271 | #elif ARCH_ARM | ||
| 272 | ff_mpegvideoencdsp_init_arm(c, avctx); | ||
| 273 | #elif ARCH_PPC | ||
| 274 | ff_mpegvideoencdsp_init_ppc(c, avctx); | ||
| 275 | #elif ARCH_RISCV | ||
| 276 | ff_mpegvideoencdsp_init_riscv(c, avctx); | ||
| 277 | #elif ARCH_X86 | ||
| 278 | 375 | ff_mpegvideoencdsp_init_x86(c, avctx); | |
| 279 | #elif ARCH_MIPS | ||
| 280 | ff_mpegvideoencdsp_init_mips(c, avctx); | ||
| 281 | #endif | ||
| 282 | 375 | } | |
| 283 |