2 * VIS optimized software YUV to RGB converter
3 * Copyright (c) 2007 Denes Balatoni <dbalatoni@programozo.hu>
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 #include "libavutil/attributes.h"
26 #include "libswscale/swscale.h"
27 #include "libswscale/swscale_internal.h"
29 #define YUV2RGB_INIT \
30 "wr %%g0, 0x10, %%gsr \n\t" \
31 "ldd [%5], %%f32 \n\t" \
32 "ldd [%5 + 8], %%f34 \n\t" \
33 "ldd [%5 + 16], %%f36 \n\t" \
34 "ldd [%5 + 24], %%f38 \n\t" \
35 "ldd [%5 + 32], %%f40 \n\t" \
36 "ldd [%5 + 40], %%f42 \n\t" \
37 "ldd [%5 + 48], %%f44 \n\t" \
38 "ldd [%5 + 56], %%f46 \n\t" \
39 "ldd [%5 + 64], %%f48 \n\t" \
40 "ldd [%5 + 72], %%f50 \n\t"
42 #define YUV2RGB_KERNEL \
43 /* ^^^^ f0=Y f3=u f5=v */ \
44 "fmul8x16 %%f3, %%f48, %%f6 \n\t" \
45 "fmul8x16 %%f19, %%f48, %%f22 \n\t" \
46 "fmul8x16 %%f5, %%f44, %%f8 \n\t" \
47 "fmul8x16 %%f21, %%f44, %%f24 \n\t" \
48 "fmul8x16 %%f0, %%f42, %%f0 \n\t" \
49 "fmul8x16 %%f16, %%f42, %%f16 \n\t" \
50 "fmul8x16 %%f3, %%f50, %%f2 \n\t" \
51 "fmul8x16 %%f19, %%f50, %%f18 \n\t" \
52 "fmul8x16 %%f5, %%f46, %%f4 \n\t" \
53 "fmul8x16 %%f21, %%f46, %%f20 \n\t" \
55 "fpsub16 %%f6, %%f34, %%f6 \n\t" /* 1 */ \
56 "fpsub16 %%f22, %%f34, %%f22 \n\t" /* 1 */ \
57 "fpsub16 %%f8, %%f38, %%f8 \n\t" /* 3 */ \
58 "fpsub16 %%f24, %%f38, %%f24 \n\t" /* 3 */ \
59 "fpsub16 %%f0, %%f32, %%f0 \n\t" /* 0 */ \
60 "fpsub16 %%f16, %%f32, %%f16 \n\t" /* 0 */ \
61 "fpsub16 %%f2, %%f36, %%f2 \n\t" /* 2 */ \
62 "fpsub16 %%f18, %%f36, %%f18 \n\t" /* 2 */ \
63 "fpsub16 %%f4, %%f40, %%f4 \n\t" /* 4 */ \
64 "fpsub16 %%f20, %%f40, %%f20 \n\t" /* 4 */ \
66 "fpadd16 %%f0, %%f8, %%f8 \n\t" /* Gt */ \
67 "fpadd16 %%f16, %%f24, %%f24 \n\t" /* Gt */ \
68 "fpadd16 %%f0, %%f4, %%f4 \n\t" /* R */ \
69 "fpadd16 %%f16, %%f20, %%f20 \n\t" /* R */ \
70 "fpadd16 %%f0, %%f6, %%f6 \n\t" /* B */ \
71 "fpadd16 %%f16, %%f22, %%f22 \n\t" /* B */ \
72 "fpadd16 %%f8, %%f2, %%f2 \n\t" /* G */ \
73 "fpadd16 %%f24, %%f18, %%f18 \n\t" /* G */ \
75 "fpack16 %%f4, %%f4 \n\t" \
76 "fpack16 %%f20, %%f20 \n\t" \
77 "fpack16 %%f6, %%f6 \n\t" \
78 "fpack16 %%f22, %%f22 \n\t" \
79 "fpack16 %%f2, %%f2 \n\t" \
80 "fpack16 %%f18, %%f18 \n\t"
82 // FIXME: must be changed to set alpha to 255 instead of 0
83 static int vis_420P_ARGB32(SwsContext *c, uint8_t *src[], int srcStride[],
84 int srcSliceY, int srcSliceH,
85 uint8_t *dst[], int dstStride[])
87 int y, out1, out2, out3, out4, out5, out6;
89 for (y = 0; y < srcSliceH; ++y)
92 "wr %%g0, 0xd2, %%asi \n\t" /* ASI_FL16_P */
94 "ldda [%1] %%asi, %%f2 \n\t"
95 "ldda [%1 + 2] %%asi, %%f18 \n\t"
96 "ldda [%2] %%asi, %%f4 \n\t"
97 "ldda [%2 + 2] %%asi, %%f20 \n\t"
99 "ld [%0+4], %%f16 \n\t"
100 "fpmerge %%f3, %%f3, %%f2 \n\t"
101 "fpmerge %%f19, %%f19, %%f18 \n\t"
102 "fpmerge %%f5, %%f5, %%f4 \n\t"
103 "fpmerge %%f21, %%f21, %%f20 \n\t"
106 "fpmerge %%f4, %%f6, %%f8 \n\t" // r, b, t1
107 "fpmerge %%f20, %%f22, %%f24 \n\t" // r, b, t1
108 "fpmerge %%f0, %%f2, %%f10 \n\t" // 0, g, t2
109 "fpmerge %%f0, %%f18, %%f26 \n\t" // 0, g, t2
110 "fpmerge %%f10, %%f8, %%f4 \n\t" // t2, t1, msb
111 "fpmerge %%f26, %%f24, %%f20 \n\t" // t2, t1, msb
112 "fpmerge %%f11, %%f9, %%f6 \n\t" // t2, t1, lsb
113 "fpmerge %%f27, %%f25, %%f22 \n\t" // t2, t1, lsb
114 "std %%f4, [%3] \n\t"
115 "std %%f20, [%3 + 16] \n\t"
116 "std %%f6, [%3 + 8] \n\t"
117 "std %%f22, [%3 + 24] \n\t"
122 "subcc %4, 8, %4 \n\t"
124 "add %3, 32, %3 \n\t" // delay slot
125 : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
126 : "0" (src[0] + (y + srcSliceY) * srcStride[0]), "1" (src[1] + ((y + srcSliceY) >> 1) * srcStride[1]),
127 "2" (src[2] + ((y + srcSliceY) >> 1) * srcStride[2]), "3" (dst[0] + (y + srcSliceY) * dstStride[0]),
129 "5" (c->sparc_coeffs)
135 // FIXME: must be changed to set alpha to 255 instead of 0
136 static int vis_422P_ARGB32(SwsContext *c, uint8_t *src[], int srcStride[],
137 int srcSliceY, int srcSliceH,
138 uint8_t *dst[], int dstStride[])
140 int y, out1, out2, out3, out4, out5, out6;
142 for (y = 0; y < srcSliceH; ++y)
145 "wr %%g0, 0xd2, %%asi \n\t" /* ASI_FL16_P */
147 "ldda [%1] %%asi, %%f2 \n\t"
148 "ldda [%1 + 2] %%asi, %%f18 \n\t"
149 "ldda [%2] %%asi, %%f4 \n\t"
150 "ldda [%2 + 2] %%asi, %%f20 \n\t"
152 "ld [%0 + 4], %%f16 \n\t"
153 "fpmerge %%f3, %%f3, %%f2 \n\t"
154 "fpmerge %%f19, %%f19, %%f18 \n\t"
155 "fpmerge %%f5, %%f5, %%f4 \n\t"
156 "fpmerge %%f21, %%f21, %%f20 \n\t"
159 "fpmerge %%f4, %%f6, %%f8 \n\t" // r,b,t1
160 "fpmerge %%f20, %%f22, %%f24 \n\t" // r,b,t1
161 "fpmerge %%f0, %%f2, %%f10 \n\t" // 0,g,t2
162 "fpmerge %%f0, %%f18, %%f26 \n\t" // 0,g,t2
163 "fpmerge %%f10, %%f8, %%f4 \n\t" // t2,t1,msb
164 "fpmerge %%f26, %%f24, %%f20 \n\t" // t2,t1,msb
165 "fpmerge %%f11, %%f9, %%f6 \n\t" // t2,t1,lsb
166 "fpmerge %%f27, %%f25, %%f22 \n\t" // t2,t1,lsb
167 "std %%f4, [%3] \n\t"
168 "std %%f20, [%3 + 16] \n\t"
169 "std %%f6, [%3 + 8] \n\t"
170 "std %%f22, [%3 + 24] \n\t"
175 "subcc %4, 8, %4 \n\t"
177 "add %3, 32, %3 \n\t" //delay slot
178 : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
179 : "0" (src[0] + (y + srcSliceY) * srcStride[0]), "1" (src[1] + (y + srcSliceY) * srcStride[1]),
180 "2" (src[2] + (y + srcSliceY) * srcStride[2]), "3" (dst[0] + (y + srcSliceY) * dstStride[0]),
182 "5" (c->sparc_coeffs)
188 av_cold SwsFunc ff_yuv2rgb_init_vis(SwsContext *c)
190 c->sparc_coeffs[5] = c->yCoeff;
191 c->sparc_coeffs[6] = c->vgCoeff;
192 c->sparc_coeffs[7] = c->vrCoeff;
193 c->sparc_coeffs[8] = c->ubCoeff;
194 c->sparc_coeffs[9] = c->ugCoeff;
196 c->sparc_coeffs[0] = (((int16_t)c->yOffset * (int16_t)c->yCoeff >> 11) & 0xffff) * 0x0001000100010001ULL;
197 c->sparc_coeffs[1] = (((int16_t)c->uOffset * (int16_t)c->ubCoeff >> 11) & 0xffff) * 0x0001000100010001ULL;
198 c->sparc_coeffs[2] = (((int16_t)c->uOffset * (int16_t)c->ugCoeff >> 11) & 0xffff) * 0x0001000100010001ULL;
199 c->sparc_coeffs[3] = (((int16_t)c->vOffset * (int16_t)c->vgCoeff >> 11) & 0xffff) * 0x0001000100010001ULL;
200 c->sparc_coeffs[4] = (((int16_t)c->vOffset * (int16_t)c->vrCoeff >> 11) & 0xffff) * 0x0001000100010001ULL;
202 if (c->dstFormat == AV_PIX_FMT_RGB32 && c->srcFormat == AV_PIX_FMT_YUV422P && (c->dstW & 7) == 0) {
203 av_log(c, AV_LOG_INFO,
204 "SPARC VIS accelerated YUV422P -> RGB32 (WARNING: alpha value is wrong)\n");
205 return vis_422P_ARGB32;
206 } else if (c->dstFormat == AV_PIX_FMT_RGB32 && c->srcFormat == AV_PIX_FMT_YUV420P && (c->dstW & 7) == 0) {
207 av_log(c, AV_LOG_INFO,
208 "SPARC VIS accelerated YUV420P -> RGB32 (WARNING: alpha value is wrong)\n");
209 return vis_420P_ARGB32;