For bit depths <= 14, the result is saturated to 15 bits. For bit depths > 14, the result is saturated to 19 bits. x86_64: chrRangeFromJpeg8_1920_c: 2126.5 2127.4 (1.00x) chrRangeFromJpeg16_1920_c: 2331.4 2325.2 (1.00x) chrRangeToJpeg8_1920_c: 3163.0 3166.9 (1.00x) chrRangeToJpeg16_1920_c: 3163.7 2152.4 (1.47x) lumRangeFromJpeg8_1920_c: 1262.2 1263.0 (1.00x) lumRangeFromJpeg16_1920_c: 1079.5 1080.5 (1.00x) lumRangeToJpeg8_1920_c: 1860.5 1886.8 (0.99x) lumRangeToJpeg16_1920_c: 1910.2 1077.0 (1.77x) aarch64 A55: chrRangeFromJpeg8_1920_c: 28836.2 28835.2 (1.00x) chrRangeFromJpeg16_1920_c: 28840.1 28839.8 (1.00x) chrRangeToJpeg8_1920_c: 44196.2 23074.7 (1.92x) chrRangeToJpeg16_1920_c: 36527.3 17318.9 (2.11x) lumRangeFromJpeg8_1920_c: 15388.5 15389.7 (1.00x) lumRangeFromJpeg16_1920_c: 15389.3 15388.2 (1.00x) lumRangeToJpeg8_1920_c: 23069.7 19227.8 (1.20x) lumRangeToJpeg16_1920_c: 19227.8 15387.0 (1.25x) aarch64 A76: chrRangeFromJpeg8_1920_c: 6334.7 6324.4 (1.00x) chrRangeFromJpeg16_1920_c: 6336.0 6339.9 (1.00x) chrRangeToJpeg8_1920_c: 11474.5 9656.0 (1.19x) chrRangeToJpeg16_1920_c: 9640.5 6340.4 (1.52x) lumRangeFromJpeg8_1920_c: 4453.2 4422.0 (1.01x) lumRangeFromJpeg16_1920_c: 4414.2 4420.9 (1.00x) lumRangeToJpeg8_1920_c: 6645.0 5949.1 (1.12x) lumRangeToJpeg16_1920_c: 6005.2 4446.8 (1.35x)
Unified · Split
| @@ -225,6 +225,10 @@ void ff_chrRangeToJpeg_neon(int16_t *dstU, int16_t *dstV, int width); · 224 unmodified lines | |||
| 225 | 225 | ||
| 226 | av_cold void ff_sws_init_range_convert_aarch64(SwsInternal *c) | 226 | av_cold void ff_sws_init_range_convert_aarch64(SwsInternal *c) |
| 227 | { | 227 | { |
| 228 | /* This code is currently disabled because of changes in the base | ||
| 229 | * implementation of these functions. This code should be enabled | ||
| 230 | * again once those changes are ported to this architecture. */ | ||
| 231 | #if 0 | ||
| 228 | int cpu_flags = av_get_cpu_flags(); | 232 | int cpu_flags = av_get_cpu_flags(); |
| 229 | 233 | ||
| 230 | if (have_neon(cpu_flags)) { | 234 | if (have_neon(cpu_flags)) { |
| @@ -238,6 +242,7 @@ av_cold void ff_sws_init_range_convert_aarch64(SwsInternal *c) · 7 unmodified lines | |||
| 238 | } | 242 | } |
| 239 | } | 243 | } |
| 240 | } | 244 | } |
| 245 | #endif | ||
| 241 | } | 246 | } |
| 242 | 247 | ||
| 243 | av_cold void ff_sws_init_swscale_aarch64(SwsInternal *c) | 248 | av_cold void ff_sws_init_swscale_aarch64(SwsInternal *c) |
| @@ -26,6 +26,10 @@ · 25 unmodified lines | |||
| 26 | 26 | ||
| 27 | av_cold void ff_sws_init_range_convert_loongarch(SwsInternal *c) | 27 | av_cold void ff_sws_init_range_convert_loongarch(SwsInternal *c) |
| 28 | { | 28 | { |
| 29 | /* This code is currently disabled because of changes in the base | ||
| 30 | * implementation of these functions. This code should be enabled | ||
| 31 | * again once those changes are ported to this architecture. */ | ||
| 32 | #if 0 | ||
| 29 | int cpu_flags = av_get_cpu_flags(); | 33 | int cpu_flags = av_get_cpu_flags(); |
| 30 | 34 | ||
| 31 | if (have_lsx(cpu_flags)) { | 35 | if (have_lsx(cpu_flags)) { |
| @@ -52,6 +56,7 @@ av_cold void ff_sws_init_range_convert_loongarch(SwsInternal *c) · 20 unmodified lines | |||
| 52 | } | 56 | } |
| 53 | } | 57 | } |
| 54 | #endif // #if HAVE_LASX | 58 | #endif // #if HAVE_LASX |
| 59 | #endif | ||
| 55 | } | 60 | } |
| 56 | 61 | ||
| 57 | av_cold void ff_sws_init_swscale_loongarch(SwsInternal *c) | 62 | av_cold void ff_sws_init_swscale_loongarch(SwsInternal *c) |
| @@ -28,6 +28,10 @@ void ff_range_chr_from_jpeg_16_rvv(int16_t *, int16_t *, int); · 27 unmodified lines | |||
| 28 | 28 | ||
| 29 | av_cold void ff_sws_init_range_convert_riscv(SwsInternal *c) | 29 | av_cold void ff_sws_init_range_convert_riscv(SwsInternal *c) |
| 30 | { | 30 | { |
| 31 | /* This code is currently disabled because of changes in the base | ||
| 32 | * implementation of these functions. This code should be enabled | ||
| 33 | * again once those changes are ported to this architecture. */ | ||
| 34 | #if 0 | ||
| 31 | #if HAVE_RVV | 35 | #if HAVE_RVV |
| 32 | int flags = av_get_cpu_flags(); | 36 | int flags = av_get_cpu_flags(); |
| 33 | 37 | ||
| @@ -47,6 +51,7 @@ av_cold void ff_sws_init_range_convert_riscv(SwsInternal *c) · 13 unmodified lines | |||
| 47 | c->chrConvertRange = convs[from].chr; | 51 | c->chrConvertRange = convs[from].chr; |
| 48 | } | 52 | } |
| 49 | #endif | 53 | #endif |
| 54 | #endif | ||
| 50 | } | 55 | } |
| 51 | 56 | ||
| 52 | #define RVV_INPUT(name) \ | 57 | #define RVV_INPUT(name) \ |
| @@ -160,8 +160,10 @@ static void chrRangeToJpeg_c(int16_t *dstU, int16_t *dstV, int width) · 159 unmodified lines | |||
| 160 | { | 160 | { |
| 161 | int i; | 161 | int i; |
| 162 | for (i = 0; i < width; i++) { | 162 | for (i = 0; i < width; i++) { |
| 163 | dstU[i] = (FFMIN(dstU[i], 30775) * 4663 - 9289992) >> 12; // -264 | 163 | int U = (dstU[i] * 4663 - 9289992) >> 12; // -264 |
| 164 | dstV[i] = (FFMIN(dstV[i], 30775) * 4663 - 9289992) >> 12; // -264 | 164 | int V = (dstV[i] * 4663 - 9289992) >> 12; // -264 |
| 165 | dstU[i] = FFMIN(U, (1 << 15) - 1); | ||
| 166 | dstV[i] = FFMIN(V, (1 << 15) - 1); | ||
| 165 | } | 167 | } |
| 166 | } | 168 | } |
| 167 | 169 | ||
| @@ -177,8 +179,10 @@ static void chrRangeFromJpeg_c(int16_t *dstU, int16_t *dstV, int width) · 9 unmodified lines | |||
| 177 | static void lumRangeToJpeg_c(int16_t *dst, int width) | 179 | static void lumRangeToJpeg_c(int16_t *dst, int width) |
| 178 | { | 180 | { |
| 179 | int i; | 181 | int i; |
| 180 | for (i = 0; i < width; i++) | 182 | for (i = 0; i < width; i++) { |
| 181 | dst[i] = (FFMIN(dst[i], 30189) * 19077 - 39057361) >> 14; | 183 | int Y = (dst[i] * 19077 - 39057361) >> 14; |
| 184 | dst[i] = FFMIN(Y, (1 << 15) - 1); | ||
| 185 | } | ||
| 182 | } | 186 | } |
| 183 | 187 | ||
| 184 | static void lumRangeFromJpeg_c(int16_t *dst, int width) | 188 | static void lumRangeFromJpeg_c(int16_t *dst, int width) |
| @@ -194,8 +198,10 @@ static void chrRangeToJpeg16_c(int16_t *_dstU, int16_t *_dstV, int width) · 9 unmodified lines | |||
| 194 | int32_t *dstU = (int32_t *) _dstU; | 198 | int32_t *dstU = (int32_t *) _dstU; |
| 195 | int32_t *dstV = (int32_t *) _dstV; | 199 | int32_t *dstV = (int32_t *) _dstV; |
| 196 | for (i = 0; i < width; i++) { | 200 | for (i = 0; i < width; i++) { |
| 197 | dstU[i] = ((int)(FFMIN(dstU[i], 30775 << 4) * 4663U - (9289992 << 4))) >> 12; // -264 | 201 | int U = ((int)(dstU[i] * 4663U - (9289992 << 4))) >> 12; // -264 |
| 198 | dstV[i] = ((int)(FFMIN(dstV[i], 30775 << 4) * 4663U - (9289992 << 4))) >> 12; // -264 | 202 | int V = ((int)(dstV[i] * 4663U - (9289992 << 4))) >> 12; // -264 |
| 203 | dstU[i] = FFMIN(U, (1 << 19) - 1); | ||
| 204 | dstV[i] = FFMIN(V, (1 << 19) - 1); | ||
| 199 | } | 205 | } |
| 200 | } | 206 | } |
| 201 | 207 | ||
| @@ -215,7 +221,8 @@ static void lumRangeToJpeg16_c(int16_t *_dst, int width) · 13 unmodified lines | |||
| 215 | int i; | 221 | int i; |
| 216 | int32_t *dst = (int32_t *) _dst; | 222 | int32_t *dst = (int32_t *) _dst; |
| 217 | for (i = 0; i < width; i++) { | 223 | for (i = 0; i < width; i++) { |
| 218 | dst[i] = ((int)(FFMIN(dst[i], 30189 << 4) * 4769U - (39057361 << 2))) >> 12; | 224 | int Y = ((int)(dst[i] * 4769U - (39057361 << 2))) >> 12; |
| 225 | dst[i] = FFMIN(Y, (1 << 19) - 1); | ||
| 219 | } | 226 | } |
| 220 | } | 227 | } |
| 221 | 228 | ||
| @@ -40,9 +40,6 @@ lum_from_offset: times 4 dd 33561947 · 39 unmodified lines | |||
| 40 | 40 | ||
| 41 | SECTION .text | 41 | SECTION .text |
| 42 | 42 | ||
| 43 | ; NOTE: there is no need to clamp the input when converting to jpeg range | ||
| 44 | ; (like we do in the C code) because packssdw will saturate the output. | ||
| 45 | |||
| 46 | ;----------------------------------------------------------------------------- | 43 | ;----------------------------------------------------------------------------- |
| 47 | ; lumConvertRange | 44 | ; lumConvertRange |
| 48 | ; | 45 | ; |