Browse code
swscale/arm/yuv2rgb: remove 32bit code path
Matthieu Bouron authored on 2016/03/23 01:46:10Showing 2 changed files
| ... | ... |
@@ -61,14 +61,14 @@ static int rgbx_to_nv12_neon_16_wrapper(SwsContext *context, const uint8_t *src[ |
| 61 | 61 |
return 0; |
| 62 | 62 |
} |
| 63 | 63 |
|
| 64 |
-#define YUV_TO_RGB_TABLE(precision) \ |
|
| 65 |
- c->yuv2rgb_v2r_coeff / ((precision) == 16 ? 1 << 7 : 1), \ |
|
| 66 |
- c->yuv2rgb_u2g_coeff / ((precision) == 16 ? 1 << 7 : 1), \ |
|
| 67 |
- c->yuv2rgb_v2g_coeff / ((precision) == 16 ? 1 << 7 : 1), \ |
|
| 68 |
- c->yuv2rgb_u2b_coeff / ((precision) == 16 ? 1 << 7 : 1), \ |
|
| 69 |
- |
|
| 70 |
-#define DECLARE_FF_YUVX_TO_RGBX_FUNCS(ifmt, ofmt, precision) \ |
|
| 71 |
-int ff_##ifmt##_to_##ofmt##_neon_##precision(int w, int h, \ |
|
| 64 |
+#define YUV_TO_RGB_TABLE \ |
|
| 65 |
+ c->yuv2rgb_v2r_coeff / (1 << 7), \ |
|
| 66 |
+ c->yuv2rgb_u2g_coeff / (1 << 7), \ |
|
| 67 |
+ c->yuv2rgb_v2g_coeff / (1 << 7), \ |
|
| 68 |
+ c->yuv2rgb_u2b_coeff / (1 << 7), \ |
|
| 69 |
+ |
|
| 70 |
+#define DECLARE_FF_YUVX_TO_RGBX_FUNCS(ifmt, ofmt) \ |
|
| 71 |
+int ff_##ifmt##_to_##ofmt##_neon(int w, int h, \ |
|
| 72 | 72 |
uint8_t *dst, int linesize, \ |
| 73 | 73 |
const uint8_t *srcY, int linesizeY, \ |
| 74 | 74 |
const uint8_t *srcU, int linesizeU, \ |
| ... | ... |
@@ -77,37 +77,34 @@ int ff_##ifmt##_to_##ofmt##_neon_##precision(int w, int h, |
| 77 | 77 |
int y_offset, \ |
| 78 | 78 |
int y_coeff); \ |
| 79 | 79 |
\ |
| 80 |
-static int ifmt##_to_##ofmt##_neon_wrapper_##precision(SwsContext *c, const uint8_t *src[], \ |
|
| 80 |
+static int ifmt##_to_##ofmt##_neon_wrapper(SwsContext *c, const uint8_t *src[], \ |
|
| 81 | 81 |
int srcStride[], int srcSliceY, int srcSliceH, \ |
| 82 | 82 |
uint8_t *dst[], int dstStride[]) { \
|
| 83 |
- const int16_t yuv2rgb_table[] = { YUV_TO_RGB_TABLE(precision) }; \
|
|
| 83 |
+ const int16_t yuv2rgb_table[] = { YUV_TO_RGB_TABLE }; \
|
|
| 84 | 84 |
\ |
| 85 |
- ff_##ifmt##_to_##ofmt##_neon_##precision(c->srcW, srcSliceH, \ |
|
| 85 |
+ ff_##ifmt##_to_##ofmt##_neon(c->srcW, srcSliceH, \ |
|
| 86 | 86 |
dst[0] + srcSliceY * dstStride[0], dstStride[0], \ |
| 87 | 87 |
src[0], srcStride[0], \ |
| 88 | 88 |
src[1], srcStride[1], \ |
| 89 | 89 |
src[2], srcStride[2], \ |
| 90 | 90 |
yuv2rgb_table, \ |
| 91 | 91 |
c->yuv2rgb_y_offset >> 9, \ |
| 92 |
- c->yuv2rgb_y_coeff / ((precision) == 16 ? 1 << 7 : 1)); \ |
|
| 92 |
+ c->yuv2rgb_y_coeff / (1 << 7)); \ |
|
| 93 | 93 |
\ |
| 94 | 94 |
return 0; \ |
| 95 | 95 |
} \ |
| 96 | 96 |
|
| 97 |
-#define DECLARE_FF_YUVX_TO_ALL_RGBX_FUNCS(yuvx, precision) \ |
|
| 98 |
-DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, argb, precision) \ |
|
| 99 |
-DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, rgba, precision) \ |
|
| 100 |
-DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, abgr, precision) \ |
|
| 101 |
-DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, bgra, precision) \ |
|
| 97 |
+#define DECLARE_FF_YUVX_TO_ALL_RGBX_FUNCS(yuvx) \ |
|
| 98 |
+DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, argb) \ |
|
| 99 |
+DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, rgba) \ |
|
| 100 |
+DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, abgr) \ |
|
| 101 |
+DECLARE_FF_YUVX_TO_RGBX_FUNCS(yuvx, bgra) \ |
|
| 102 | 102 |
|
| 103 |
-#define DECLARE_FF_YUVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(yuvx) \ |
|
| 104 |
-DECLARE_FF_YUVX_TO_ALL_RGBX_FUNCS(yuvx, 16) \ |
|
| 103 |
+DECLARE_FF_YUVX_TO_ALL_RGBX_FUNCS(yuv420p) |
|
| 104 |
+DECLARE_FF_YUVX_TO_ALL_RGBX_FUNCS(yuv422p) |
|
| 105 | 105 |
|
| 106 |
-DECLARE_FF_YUVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(yuv420p) |
|
| 107 |
-DECLARE_FF_YUVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(yuv422p) |
|
| 108 |
- |
|
| 109 |
-#define DECLARE_FF_NVX_TO_RGBX_FUNCS(ifmt, ofmt, precision) \ |
|
| 110 |
-int ff_##ifmt##_to_##ofmt##_neon_##precision(int w, int h, \ |
|
| 106 |
+#define DECLARE_FF_NVX_TO_RGBX_FUNCS(ifmt, ofmt) \ |
|
| 107 |
+int ff_##ifmt##_to_##ofmt##_neon(int w, int h, \ |
|
| 111 | 108 |
uint8_t *dst, int linesize, \ |
| 112 | 109 |
const uint8_t *srcY, int linesizeY, \ |
| 113 | 110 |
const uint8_t *srcC, int linesizeC, \ |
| ... | ... |
@@ -115,32 +112,29 @@ int ff_##ifmt##_to_##ofmt##_neon_##precision(int w, int h, |
| 115 | 115 |
int y_offset, \ |
| 116 | 116 |
int y_coeff); \ |
| 117 | 117 |
\ |
| 118 |
-static int ifmt##_to_##ofmt##_neon_wrapper_##precision(SwsContext *c, const uint8_t *src[], \ |
|
| 118 |
+static int ifmt##_to_##ofmt##_neon_wrapper(SwsContext *c, const uint8_t *src[], \ |
|
| 119 | 119 |
int srcStride[], int srcSliceY, int srcSliceH, \ |
| 120 | 120 |
uint8_t *dst[], int dstStride[]) { \
|
| 121 |
- const int16_t yuv2rgb_table[] = { YUV_TO_RGB_TABLE(precision) }; \
|
|
| 121 |
+ const int16_t yuv2rgb_table[] = { YUV_TO_RGB_TABLE }; \
|
|
| 122 | 122 |
\ |
| 123 |
- ff_##ifmt##_to_##ofmt##_neon_##precision(c->srcW, srcSliceH, \ |
|
| 123 |
+ ff_##ifmt##_to_##ofmt##_neon(c->srcW, srcSliceH, \ |
|
| 124 | 124 |
dst[0] + srcSliceY * dstStride[0], dstStride[0], \ |
| 125 | 125 |
src[0], srcStride[0], src[1], srcStride[1], \ |
| 126 | 126 |
yuv2rgb_table, \ |
| 127 | 127 |
c->yuv2rgb_y_offset >> 9, \ |
| 128 |
- c->yuv2rgb_y_coeff / ((precision) == 16 ? 1 << 7 : 1)); \ |
|
| 128 |
+ c->yuv2rgb_y_coeff / (1 << 7)); \ |
|
| 129 | 129 |
\ |
| 130 | 130 |
return 0; \ |
| 131 | 131 |
} \ |
| 132 | 132 |
|
| 133 |
-#define DECLARE_FF_NVX_TO_ALL_RGBX_FUNCS(nvx, precision) \ |
|
| 134 |
-DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, argb, precision) \ |
|
| 135 |
-DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, rgba, precision) \ |
|
| 136 |
-DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, abgr, precision) \ |
|
| 137 |
-DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, bgra, precision) \ |
|
| 138 |
- |
|
| 139 |
-#define DECLARE_FF_NVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(nvx) \ |
|
| 140 |
-DECLARE_FF_NVX_TO_ALL_RGBX_FUNCS(nvx, 16) \ |
|
| 133 |
+#define DECLARE_FF_NVX_TO_ALL_RGBX_FUNCS(nvx) \ |
|
| 134 |
+DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, argb) \ |
|
| 135 |
+DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, rgba) \ |
|
| 136 |
+DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, abgr) \ |
|
| 137 |
+DECLARE_FF_NVX_TO_RGBX_FUNCS(nvx, bgra) \ |
|
| 141 | 138 |
|
| 142 |
-DECLARE_FF_NVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(nv12) |
|
| 143 |
-DECLARE_FF_NVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(nv21) |
|
| 139 |
+DECLARE_FF_NVX_TO_ALL_RGBX_FUNCS(nv12) |
|
| 140 |
+DECLARE_FF_NVX_TO_ALL_RGBX_FUNCS(nv21) |
|
| 144 | 141 |
|
| 145 | 142 |
/* We need a 16 pixel width alignment. This constraint can easily be removed |
| 146 | 143 |
* for input reading but for the output which is 4-bytes per pixel (RGBA) the |
| ... | ... |
@@ -152,7 +146,7 @@ DECLARE_FF_NVX_TO_ALL_RGBX_ALL_PRECISION_FUNCS(nv21) |
| 152 | 152 |
&& !(c->srcH & 1) \ |
| 153 | 153 |
&& !(c->srcW & 15) \ |
| 154 | 154 |
&& !accurate_rnd) { \
|
| 155 |
- c->swscale = ifmt##_to_##ofmt##_neon_wrapper_16; \ |
|
| 155 |
+ c->swscale = ifmt##_to_##ofmt##_neon_wrapper; \ |
|
| 156 | 156 |
} \ |
| 157 | 157 |
} while (0) |
| 158 | 158 |
|
| ... | ... |
@@ -22,7 +22,7 @@ |
| 22 | 22 |
#include "libavutil/arm/asm.S" |
| 23 | 23 |
|
| 24 | 24 |
|
| 25 |
-.macro compute_premult_16 half_u1, half_u2, half_v1, half_v2 |
|
| 25 |
+.macro compute_premult half_u1, half_u2, half_v1, half_v2 |
|
| 26 | 26 |
vmov d2, \half_u1 @ copy left q14 to left q1 |
| 27 | 27 |
vmov d3, \half_u1 @ copy left q14 to right q1 |
| 28 | 28 |
vmov d4, \half_u2 @ copy right q14 to left q2 |
| ... | ... |
@@ -49,56 +49,22 @@ |
| 49 | 49 |
vmul.s16 q13, q2, d1[3] @ U * u2b (right, blue) |
| 50 | 50 |
.endm |
| 51 | 51 |
|
| 52 |
-.macro compute_premult_32 half_u half_v |
|
| 53 |
- vmov d2, \half_u @ copy left q14 to left q1 |
|
| 54 |
- vmov d3, \half_u @ copy left q14 to right q1 |
|
| 55 |
- vmov d4, \half_v @ copy left q15 to left q2 |
|
| 56 |
- vmov d5, \half_v @ copy left q15 to right q2 |
|
| 57 |
- |
|
| 58 |
- vzip.16 d2, d3 @ U1U1U2U2U3U3U4U4 |
|
| 59 |
- vzip.16 d4, d5 @ V1V1V2V2V3V3V4V4 |
|
| 60 |
- |
|
| 61 |
- vmull.s16 q8, d4, d1[0] @ V * v2r (left, red) |
|
| 62 |
- vmull.s16 q9, d5, d1[0] @ V * v2r (right, red) |
|
| 63 |
- vmull.s16 q10, d2, d1[1] @ U * u2g |
|
| 64 |
- vmull.s16 q11, d3, d1[1] @ U * u2g |
|
| 65 |
- vmlal.s16 q10, d4, d1[2] @ U * u2g + V * v2g (left, green) |
|
| 66 |
- vmlal.s16 q11, d5, d1[2] @ U * u2g + V * v2g (right, green) |
|
| 67 |
- vmull.s16 q12, d2, d1[3] @ U * u2b (left, blue) |
|
| 68 |
- vmull.s16 q13, d3, d1[3] @ U * u2b (right, blue) |
|
| 69 |
-.endm |
|
| 70 |
- |
|
| 71 |
-.macro compute_color_16 dst_comp1 dst_comp2 pre1 pre2 |
|
| 52 |
+.macro compute_color dst_comp1 dst_comp2 pre1 pre2 |
|
| 72 | 53 |
vadd.s16 q1, q14, \pre1 |
| 73 | 54 |
vadd.s16 q2, q15, \pre2 |
| 74 | 55 |
vqrshrun.s16 \dst_comp1, q1, #6 |
| 75 | 56 |
vqrshrun.s16 \dst_comp2, q2, #6 |
| 76 | 57 |
.endm |
| 77 | 58 |
|
| 78 |
-.macro compute_color_32 dst_comp pre1 pre2 |
|
| 79 |
- vadd.s32 q3, q1, \pre1 |
|
| 80 |
- vadd.s32 q4, q2, \pre2 |
|
| 81 |
- vqrshrun.s32 d10, q3, #13 |
|
| 82 |
- vqrshrun.s32 d11, q4, #13 @ q5 = ({q3,q4} + (1<<12)) >> 13
|
|
| 83 |
- vqmovn.u16 \dst_comp, q5 @ saturate 16bit -> 8bit |
|
| 84 |
-.endm |
|
| 85 |
- |
|
| 86 |
-.macro compute_rgba_16 r1 r2 g1 g2 b1 b2 a1 a2 |
|
| 87 |
- compute_color_16 \r1, \r2, q8, q9 |
|
| 88 |
- compute_color_16 \g1, \g2, q10, q11 |
|
| 89 |
- compute_color_16 \b1, \b2, q12, q13 |
|
| 59 |
+.macro compute_rgba r1 r2 g1 g2 b1 b2 a1 a2 |
|
| 60 |
+ compute_color \r1, \r2, q8, q9 |
|
| 61 |
+ compute_color \g1, \g2, q10, q11 |
|
| 62 |
+ compute_color \b1, \b2, q12, q13 |
|
| 90 | 63 |
vmov.u8 \a1, #255 |
| 91 | 64 |
vmov.u8 \a2, #255 |
| 92 | 65 |
.endm |
| 93 | 66 |
|
| 94 |
-.macro compute_rgba_32 r g b a |
|
| 95 |
- compute_color_32 \r, q8, q9 |
|
| 96 |
- compute_color_32 \g, q10, q11 |
|
| 97 |
- compute_color_32 \b, q12, q13 |
|
| 98 |
- vmov.u8 \a, #255 |
|
| 99 |
-.endm |
|
| 100 |
- |
|
| 101 |
-.macro compute_16px_16 dst y0 y1 ofmt |
|
| 67 |
+.macro compute_16px dst y0 y1 ofmt |
|
| 102 | 68 |
vmovl.u8 q14, \y0 @ 8px of y |
| 103 | 69 |
vmovl.u8 q15, \y1 @ 8px of y |
| 104 | 70 |
|
| ... | ... |
@@ -114,91 +80,39 @@ |
| 114 | 114 |
|
| 115 | 115 |
|
| 116 | 116 |
.ifc \ofmt,argb |
| 117 |
- compute_rgba_16 d7, d11, d8, d12, d9, d13, d6, d10 |
|
| 117 |
+ compute_rgba d7, d11, d8, d12, d9, d13, d6, d10 |
|
| 118 | 118 |
.endif |
| 119 | 119 |
|
| 120 | 120 |
.ifc \ofmt,rgba |
| 121 |
- compute_rgba_16 d6, d10, d7, d11, d8, d12, d9, d13 |
|
| 121 |
+ compute_rgba d6, d10, d7, d11, d8, d12, d9, d13 |
|
| 122 | 122 |
.endif |
| 123 | 123 |
|
| 124 | 124 |
.ifc \ofmt,abgr |
| 125 |
- compute_rgba_16 d9, d13, d8, d12, d7, d11, d6, d10 |
|
| 125 |
+ compute_rgba d9, d13, d8, d12, d7, d11, d6, d10 |
|
| 126 | 126 |
.endif |
| 127 | 127 |
|
| 128 | 128 |
.ifc \ofmt,bgra |
| 129 |
- compute_rgba_16 d8, d12, d7, d11, d6, d10, d9, d13 |
|
| 129 |
+ compute_rgba d8, d12, d7, d11, d6, d10, d9, d13 |
|
| 130 | 130 |
.endif |
| 131 | 131 |
vst4.8 {q3, q4}, [\dst,:128]!
|
| 132 | 132 |
vst4.8 {q5, q6}, [\dst,:128]!
|
| 133 | 133 |
|
| 134 | 134 |
.endm |
| 135 | 135 |
|
| 136 |
-.macro compute_8px_32 dst half_y ofmt |
|
| 137 |
- vmovl.u8 q7, \half_y @ 8px of Y |
|
| 138 |
- vdup.16 q5, r9 |
|
| 139 |
- vsub.s16 q7, q5 |
|
| 140 |
- vmull.s16 q1, d14, d0 @ q1 = (srcY - y_offset) * y_coeff (left) |
|
| 141 |
- vmull.s16 q2, d15, d0 @ q2 = (srcY - y_offset) * y_coeff (right) |
|
| 142 |
- |
|
| 143 |
-.ifc \ofmt,argb |
|
| 144 |
- compute_rgba_32 d13, d14, d15, d12 |
|
| 145 |
-.endif |
|
| 146 |
- |
|
| 147 |
-.ifc \ofmt,rgba |
|
| 148 |
- compute_rgba_32 d12, d13, d14, d15 |
|
| 149 |
-.endif |
|
| 150 |
- |
|
| 151 |
-.ifc \ofmt,abgr |
|
| 152 |
- compute_rgba_32 d15, d14, d13, d12 |
|
| 153 |
-.endif |
|
| 154 |
- |
|
| 155 |
-.ifc \ofmt,bgra |
|
| 156 |
- compute_rgba_32 d14, d13, d12, d15 |
|
| 157 |
-.endif |
|
| 158 |
- |
|
| 159 |
- vst4.8 {q6, q7}, [\dst,:128]!
|
|
| 160 |
-.endm |
|
| 161 |
- |
|
| 162 |
-.macro process_1l_16px_16 ofmt |
|
| 163 |
- compute_premult_16 d28, d29, d30, d31 |
|
| 136 |
+.macro process_1l_16px ofmt |
|
| 137 |
+ compute_premult d28, d29, d30, d31 |
|
| 164 | 138 |
vld1.8 {q7}, [r4]!
|
| 165 |
- compute_16px_16 r2, d14, d15, \ofmt |
|
| 139 |
+ compute_16px r2, d14, d15, \ofmt |
|
| 166 | 140 |
.endm |
| 167 | 141 |
|
| 168 |
-.macro process_1l_16px_32 ofmt |
|
| 169 |
- compute_premult_32 d28, d30 |
|
| 170 |
- vld1.8 {q7}, [r4]!
|
|
| 171 |
- vmov d28, d15 @ save right of the line of luma for later use |
|
| 172 |
- compute_8px_32 r2, d14, \ofmt |
|
| 173 |
- |
|
| 174 |
- compute_premult_32 d29, d31 |
|
| 175 |
- compute_8px_32 r2, d28, \ofmt |
|
| 176 |
-.endm |
|
| 177 |
- |
|
| 178 |
-.macro process_2l_16px_16 ofmt |
|
| 179 |
- compute_premult_16 d28, d29, d30, d31 |
|
| 180 |
- |
|
| 181 |
- vld1.8 {q7}, [r4]! @ first line of luma
|
|
| 182 |
- compute_16px_16 r2, d14, d15, \ofmt |
|
| 183 |
- |
|
| 184 |
- vld1.8 {q7}, [r12]! @ second line of luma
|
|
| 185 |
- compute_16px_16 r11, d14, d15, \ofmt |
|
| 186 |
-.endm |
|
| 187 |
- |
|
| 188 |
-.macro process_2l_16px_32 ofmt |
|
| 189 |
- compute_premult_32 d28, d30 |
|
| 142 |
+.macro process_2l_16px ofmt |
|
| 143 |
+ compute_premult d28, d29, d30, d31 |
|
| 190 | 144 |
|
| 191 | 145 |
vld1.8 {q7}, [r4]! @ first line of luma
|
| 192 |
- vmov d28, d15 @ save right of the first line of luma for later use |
|
| 193 |
- compute_8px_32 r2, d14, \ofmt |
|
| 146 |
+ compute_16px r2, d14, d15, \ofmt |
|
| 194 | 147 |
|
| 195 | 148 |
vld1.8 {q7}, [r12]! @ second line of luma
|
| 196 |
- vmov d30, d15 @ save right of the second line of luma for later use |
|
| 197 |
- compute_8px_32 r11, d14, \ofmt |
|
| 198 |
- |
|
| 199 |
- compute_premult_32 d29, d31 |
|
| 200 |
- compute_8px_32 r2, d28, \ofmt |
|
| 201 |
- compute_8px_32 r11, d30, \ofmt |
|
| 149 |
+ compute_16px r11, d14, d15, \ofmt |
|
| 202 | 150 |
.endm |
| 203 | 151 |
|
| 204 | 152 |
.macro load_args_nvx |
| ... | ... |
@@ -266,8 +180,8 @@ |
| 266 | 266 |
ldr r10,[sp, #120] @ r10 = srcV |
| 267 | 267 |
.endm |
| 268 | 268 |
|
| 269 |
-.macro declare_func ifmt ofmt precision |
|
| 270 |
-function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
|
| 269 |
+.macro declare_func ifmt ofmt |
|
| 270 |
+function ff_\ifmt\()_to_\ofmt\()_neon, export=1 |
|
| 271 | 271 |
|
| 272 | 272 |
.ifc \ifmt,nv12 |
| 273 | 273 |
load_args_nvx |
| ... | ... |
@@ -301,7 +215,7 @@ function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
| 301 | 301 |
vsubl.u8 q14, d2, d10 @ q14 = U - 128 |
| 302 | 302 |
vsubl.u8 q15, d3, d10 @ q15 = V - 128 |
| 303 | 303 |
|
| 304 |
- process_2l_16px_\precision \ofmt |
|
| 304 |
+ process_2l_16px \ofmt |
|
| 305 | 305 |
.endif |
| 306 | 306 |
|
| 307 | 307 |
.ifc \ifmt,nv21 |
| ... | ... |
@@ -311,7 +225,7 @@ function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
| 311 | 311 |
vsubl.u8 q14, d3, d10 @ q14 = U - 128 |
| 312 | 312 |
vsubl.u8 q15, d2, d10 @ q15 = V - 128 |
| 313 | 313 |
|
| 314 |
- process_2l_16px_\precision \ofmt |
|
| 314 |
+ process_2l_16px \ofmt |
|
| 315 | 315 |
.endif |
| 316 | 316 |
|
| 317 | 317 |
.ifc \ifmt,yuv420p |
| ... | ... |
@@ -323,7 +237,7 @@ function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
| 323 | 323 |
vsubl.u8 q14, d2, d10 @ q14 = U - 128 |
| 324 | 324 |
vsubl.u8 q15, d3, d10 @ q15 = V - 128 |
| 325 | 325 |
|
| 326 |
- process_2l_16px_\precision \ofmt |
|
| 326 |
+ process_2l_16px \ofmt |
|
| 327 | 327 |
.endif |
| 328 | 328 |
|
| 329 | 329 |
.ifc \ifmt,yuv422p |
| ... | ... |
@@ -334,7 +248,7 @@ function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
| 334 | 334 |
vsubl.u8 q14, d2, d10 @ q14 = U - 128 |
| 335 | 335 |
vsubl.u8 q15, d3, d10 @ q15 = V - 128 |
| 336 | 336 |
|
| 337 |
- process_1l_16px_\precision \ofmt |
|
| 337 |
+ process_1l_16px \ofmt |
|
| 338 | 338 |
.endif |
| 339 | 339 |
|
| 340 | 340 |
subs r8, r8, #16 @ width -= 16 |
| ... | ... |
@@ -390,18 +304,14 @@ function ff_\ifmt\()_to_\ofmt\()_neon_\precision\(), export=1 |
| 390 | 390 |
endfunc |
| 391 | 391 |
.endm |
| 392 | 392 |
|
| 393 |
-.macro declare_rgb_funcs ifmt precision |
|
| 394 |
- declare_func \ifmt, argb, \precision |
|
| 395 |
- declare_func \ifmt, rgba, \precision |
|
| 396 |
- declare_func \ifmt, abgr, \precision |
|
| 397 |
- declare_func \ifmt, bgra, \precision |
|
| 393 |
+.macro declare_rgb_funcs ifmt |
|
| 394 |
+ declare_func \ifmt, argb |
|
| 395 |
+ declare_func \ifmt, rgba |
|
| 396 |
+ declare_func \ifmt, abgr |
|
| 397 |
+ declare_func \ifmt, bgra |
|
| 398 | 398 |
.endm |
| 399 | 399 |
|
| 400 |
-declare_rgb_funcs nv12, 16 |
|
| 401 |
-declare_rgb_funcs nv21, 16 |
|
| 402 |
-declare_rgb_funcs nv12, 32 |
|
| 403 |
-declare_rgb_funcs nv21, 32 |
|
| 404 |
-declare_rgb_funcs yuv420p, 16 |
|
| 405 |
-declare_rgb_funcs yuv420p, 32 |
|
| 406 |
-declare_rgb_funcs yuv422p, 16 |
|
| 407 |
-declare_rgb_funcs yuv422p, 32 |
|
| 400 |
+declare_rgb_funcs nv12 |
|
| 401 |
+declare_rgb_funcs nv21 |
|
| 402 |
+declare_rgb_funcs yuv420p |
|
| 403 |
+declare_rgb_funcs yuv422p |