yuv2rgb_vis.c
Go to the documentation of this file.
1 /*
2  * VIS optimized software YUV to RGB converter
3  * Copyright (c) 2007 Denes Balatoni <dbalatoni@programozo.hu>
4  *
5  * This file is part of Libav.
6  *
7  * Libav 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.
11  *
12  * Libav 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.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with Libav; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include <inttypes.h>
23 #include <stdlib.h>
24 
25 #include "libswscale/swscale.h"
27 
28 #define YUV2RGB_INIT \
29  "wr %%g0, 0x10, %%gsr \n\t" \
30  "ldd [%5], %%f32 \n\t" \
31  "ldd [%5+8], %%f34 \n\t" \
32  "ldd [%5+16], %%f36 \n\t" \
33  "ldd [%5+24], %%f38 \n\t" \
34  "ldd [%5+32], %%f40 \n\t" \
35  "ldd [%5+40], %%f42 \n\t" \
36  "ldd [%5+48], %%f44 \n\t" \
37  "ldd [%5+56], %%f46 \n\t" \
38  "ldd [%5+64], %%f48 \n\t" \
39  "ldd [%5+72], %%f50 \n\t"
40 
41 #define YUV2RGB_KERNEL \
42  /* ^^^^ f0=Y f3=u f5=v */ \
43  "fmul8x16 %%f3, %%f48, %%f6 \n\t" \
44  "fmul8x16 %%f19, %%f48, %%f22 \n\t" \
45  "fmul8x16 %%f5, %%f44, %%f8 \n\t" \
46  "fmul8x16 %%f21, %%f44, %%f24 \n\t" \
47  "fmul8x16 %%f0, %%f42, %%f0 \n\t" \
48  "fmul8x16 %%f16, %%f42, %%f16 \n\t" \
49  "fmul8x16 %%f3, %%f50, %%f2 \n\t" \
50  "fmul8x16 %%f19, %%f50, %%f18 \n\t" \
51  "fmul8x16 %%f5, %%f46, %%f4 \n\t" \
52  "fmul8x16 %%f21, %%f46, %%f20 \n\t" \
53  \
54  "fpsub16 %%f6, %%f34, %%f6 \n\t" /* 1 */ \
55  "fpsub16 %%f22, %%f34, %%f22 \n\t" /* 1 */ \
56  "fpsub16 %%f8, %%f38, %%f8 \n\t" /* 3 */ \
57  "fpsub16 %%f24, %%f38, %%f24 \n\t" /* 3 */ \
58  "fpsub16 %%f0, %%f32, %%f0 \n\t" /* 0 */ \
59  "fpsub16 %%f16, %%f32, %%f16 \n\t" /* 0 */ \
60  "fpsub16 %%f2, %%f36, %%f2 \n\t" /* 2 */ \
61  "fpsub16 %%f18, %%f36, %%f18 \n\t" /* 2 */ \
62  "fpsub16 %%f4, %%f40, %%f4 \n\t" /* 4 */ \
63  "fpsub16 %%f20, %%f40, %%f20 \n\t" /* 4 */ \
64  \
65  "fpadd16 %%f0, %%f8, %%f8 \n\t" /* Gt */ \
66  "fpadd16 %%f16, %%f24, %%f24 \n\t" /* Gt */ \
67  "fpadd16 %%f0, %%f4, %%f4 \n\t" /* R */ \
68  "fpadd16 %%f16, %%f20, %%f20 \n\t" /* R */ \
69  "fpadd16 %%f0, %%f6, %%f6 \n\t" /* B */ \
70  "fpadd16 %%f16, %%f22, %%f22 \n\t" /* B */ \
71  "fpadd16 %%f8, %%f2, %%f2 \n\t" /* G */ \
72  "fpadd16 %%f24, %%f18, %%f18 \n\t" /* G */ \
73  \
74  "fpack16 %%f4, %%f4 \n\t" \
75  "fpack16 %%f20, %%f20 \n\t" \
76  "fpack16 %%f6, %%f6 \n\t" \
77  "fpack16 %%f22, %%f22 \n\t" \
78  "fpack16 %%f2, %%f2 \n\t" \
79  "fpack16 %%f18, %%f18 \n\t"
80 
81 
82 
83 // FIXME: must be changed to set alpha to 255 instead of 0
84 static int vis_420P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
85  int srcSliceH, uint8_t* dst[], int dstStride[])
86 {
87  int y, out1, out2, out3, out4, out5, out6;
88 
89  for(y=0;y < srcSliceH;++y) {
90  __asm__ volatile (
92  "wr %%g0, 0xd2, %%asi \n\t" /* ASI_FL16_P */
93  "1: \n\t"
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"
98  "ld [%0], %%f0 \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"
105  "fzero %%f0 \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"
118 
119  "add %0, 8, %0 \n\t"
120  "add %1, 4, %1 \n\t"
121  "add %2, 4, %2 \n\t"
122  "subcc %4, 8, %4 \n\t"
123  "bne 1b \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]),
128  "4" (c->dstW),
129  "5" (c->sparc_coeffs)
130  );
131  }
132 
133  return srcSliceH;
134 }
135 
136 // FIXME: must be changed to set alpha to 255 instead of 0
137 static int vis_422P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
138  int srcSliceH, uint8_t* dst[], int dstStride[])
139 {
140  int y, out1, out2, out3, out4, out5, out6;
141 
142  for(y=0;y < srcSliceH;++y) {
143  __asm__ volatile (
145  "wr %%g0, 0xd2, %%asi \n\t" /* ASI_FL16_P */
146  "1: \n\t"
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"
151  "ld [%0], %%f0 \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"
158  "fzero %%f0 \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"
171 
172  "add %0, 8, %0 \n\t"
173  "add %1, 4, %1 \n\t"
174  "add %2, 4, %2 \n\t"
175  "subcc %4, 8, %4 \n\t"
176  "bne 1b \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]),
181  "4" (c->dstW),
182  "5" (c->sparc_coeffs)
183  );
184  }
185 
186  return srcSliceH;
187 }
188 
190 {
191  c->sparc_coeffs[5]=c->yCoeff;
192  c->sparc_coeffs[6]=c->vgCoeff;
193  c->sparc_coeffs[7]=c->vrCoeff;
194  c->sparc_coeffs[8]=c->ubCoeff;
195  c->sparc_coeffs[9]=c->ugCoeff;
196 
197  c->sparc_coeffs[0]=(((int16_t)c->yOffset*(int16_t)c->yCoeff >>11) & 0xffff) * 0x0001000100010001ULL;
198  c->sparc_coeffs[1]=(((int16_t)c->uOffset*(int16_t)c->ubCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
199  c->sparc_coeffs[2]=(((int16_t)c->uOffset*(int16_t)c->ugCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
200  c->sparc_coeffs[3]=(((int16_t)c->vOffset*(int16_t)c->vgCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
201  c->sparc_coeffs[4]=(((int16_t)c->vOffset*(int16_t)c->vrCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
202 
203  if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV422P && (c->dstW & 7)==0) {
204  av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV422P -> RGB32 (WARNING: alpha value is wrong)\n");
205  return vis_422P_ARGB32;
206  }
207  else if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV420P && (c->dstW & 7)==0) {
208  av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV420P -> RGB32 (WARNING: alpha value is wrong)\n");
209  return vis_420P_ARGB32;
210  }
211  return NULL;
212 }