libswscale/yuv2rgb_vis.c

   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 FFmpeg.
   6  *
   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.
  11  *
  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.
  16  *
  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
  20  */
  21
  22 #include <inttypes.h>
  23 #include <stdlib.h>
  24
  25 #include "swscale.h"
  26
  27 #define YUV2RGB_INIT \
  28     "wr %%g0, 0x10, %%gsr \n\t" \
  29     "ldd [%5], %%f32      \n\t" \
  30     "ldd [%5+8], %%f34    \n\t" \
  31     "ldd [%5+16], %%f36   \n\t" \
  32     "ldd [%5+24], %%f38   \n\t" \
  33     "ldd [%5+32], %%f40   \n\t" \
  34     "ldd [%5+40], %%f42   \n\t" \
  35     "ldd [%5+48], %%f44   \n\t" \
  36     "ldd [%5+56], %%f46   \n\t" \
  37     "ldd [%5+64], %%f48   \n\t" \
  38     "ldd [%5+72], %%f50   \n\t"
  39
  40 #define YUV2RGB_KERNEL \
  41     /* ^^^^ f0=Y f3=u f5=v */ \
  42     "fmul8x16 %%f3, %%f48, %%f6   \n\t" \
  43     "fmul8x16 %%f19, %%f48, %%f22 \n\t" \
  44     "fmul8x16 %%f5, %%f44, %%f8   \n\t" \
  45     "fmul8x16 %%f21, %%f44, %%f24 \n\t" \
  46     "fmul8x16 %%f0, %%f42, %%f0   \n\t" \
  47     "fmul8x16 %%f16, %%f42, %%f16 \n\t" \
  48     "fmul8x16 %%f3, %%f50, %%f2   \n\t" \
  49     "fmul8x16 %%f19, %%f50, %%f18 \n\t" \
  50     "fmul8x16 %%f5, %%f46, %%f4   \n\t" \
  51     "fmul8x16 %%f21, %%f46, %%f20 \n\t" \
  52     \
  53     "fpsub16 %%f6, %%f34, %%f6   \n\t" /* 1 */ \
  54     "fpsub16 %%f22, %%f34, %%f22 \n\t" /* 1 */ \
  55     "fpsub16 %%f8, %%f38, %%f8   \n\t" /* 3 */ \
  56     "fpsub16 %%f24, %%f38, %%f24 \n\t" /* 3 */ \
  57     "fpsub16 %%f0, %%f32, %%f0   \n\t" /* 0 */ \
  58     "fpsub16 %%f16, %%f32, %%f16 \n\t" /* 0 */ \
  59     "fpsub16 %%f2, %%f36, %%f2   \n\t" /* 2 */ \
  60     "fpsub16 %%f18, %%f36, %%f18 \n\t" /* 2 */ \
  61     "fpsub16 %%f4, %%f40, %%f4   \n\t" /* 4 */ \
  62     "fpsub16 %%f20, %%f40, %%f20 \n\t" /* 4 */ \
  63     \
  64     "fpadd16 %%f0, %%f8, %%f8    \n\t" /* Gt */ \
  65     "fpadd16 %%f16, %%f24, %%f24 \n\t" /* Gt */ \
  66     "fpadd16 %%f0, %%f4, %%f4    \n\t" /* R */ \
  67     "fpadd16 %%f16, %%f20, %%f20 \n\t" /* R */ \
  68     "fpadd16 %%f0, %%f6, %%f6    \n\t" /* B */ \
  69     "fpadd16 %%f16, %%f22, %%f22 \n\t" /* B */ \
  70     "fpadd16 %%f8, %%f2, %%f2    \n\t" /* G */ \
  71     "fpadd16 %%f24, %%f18, %%f18 \n\t" /* G */ \
  72     \
  73     "fpack16 %%f4, %%f4    \n\t" \
  74     "fpack16 %%f20, %%f20  \n\t" \
  75     "fpack16 %%f6, %%f6    \n\t" \
  76     "fpack16 %%f22, %%f22  \n\t" \
  77     "fpack16 %%f2, %%f2    \n\t" \
  78     "fpack16 %%f18, %%f18  \n\t"
  79
  80
  81
  82 static int vis_420P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
  83                            int srcSliceH, uint8_t* dst[], int dstStride[]){
  84   int y, out1, out2, out3, out4, out5, out6;
  85
  86   for(y=0;y < srcSliceH;++y) {
  87       asm volatile (
  88           YUV2RGB_INIT
  89           "wr %%g0, 0xd2, %%asi        \n\t" /* ASI_FL16_P */
  90           "1:                          \n\t"
  91           "ldda [%1] %%asi, %%f2       \n\t"
  92           "ldda [%1+2] %%asi, %%f18    \n\t"
  93           "ldda [%2] %%asi, %%f4       \n\t"
  94           "ldda [%2+2] %%asi, %%f20    \n\t"
  95           "ld [%0], %%f0               \n\t"
  96           "ld [%0+4], %%f16            \n\t"
  97           "fpmerge %%f3, %%f3, %%f2    \n\t"
  98           "fpmerge %%f19, %%f19, %%f18 \n\t"
  99           "fpmerge %%f5, %%f5, %%f4    \n\t"
 100           "fpmerge %%f21, %%f21, %%f20 \n\t"
 101           YUV2RGB_KERNEL
 102           "fzero %%f0                  \n\t"
 103           "fpmerge %%f4, %%f6, %%f8    \n\t"  // r,b,t1
 104           "fpmerge %%f20, %%f22, %%f24 \n\t"  // r,b,t1
 105           "fpmerge %%f0, %%f2, %%f10   \n\t"  // 0,g,t2
 106           "fpmerge %%f0, %%f18, %%f26  \n\t"  // 0,g,t2
 107           "fpmerge %%f10, %%f8, %%f4   \n\t"  // t2,t1,msb
 108           "fpmerge %%f26, %%f24, %%f20 \n\t"  // t2,t1,msb
 109           "fpmerge %%f11, %%f9, %%f6   \n\t"  // t2,t1,lsb
 110           "fpmerge %%f27, %%f25, %%f22 \n\t"  // t2,t1,lsb
 111           "std %%f4, [%3]              \n\t"
 112           "std %%f20, [%3+16]          \n\t"
 113           "std %%f6, [%3+8]            \n\t"
 114           "std %%f22, [%3+24]          \n\t"
 115
 116           "add %0, 8, %0   \n\t"
 117           "add %1, 4, %1   \n\t"
 118           "add %2, 4, %2   \n\t"
 119           "subcc %4, 8, %4 \n\t"
 120           "bne 1b          \n\t"
 121           "add %3, 32, %3  \n\t" //delay slot
 122           : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
 123           : "0" (src[0]+(y+srcSliceY)*srcStride[0]), "1" (src[1]+((y+srcSliceY)>>1)*srcStride[1]),
 124             "2" (src[2]+((y+srcSliceY)>>1)*srcStride[2]), "3" (dst[0]+(y+srcSliceY)*dstStride[0]),
 125             "4" (c->dstW),
 126             "5" (c->sparc_coeffs)
 127       );
 128   }
 129
 130   return srcSliceH;
 131 }
 132
 133 static int vis_422P_ARGB32(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
 134                            int srcSliceH, uint8_t* dst[], int dstStride[]){
 135   int y, out1, out2, out3, out4, out5, out6;
 136
 137   for(y=0;y < srcSliceH;++y) {
 138       asm volatile (
 139           YUV2RGB_INIT
 140           "wr %%g0, 0xd2, %%asi        \n\t" /* ASI_FL16_P */
 141           "1:                          \n\t"
 142           "ldda [%1] %%asi, %%f2       \n\t"
 143           "ldda [%1+2] %%asi, %%f18    \n\t"
 144           "ldda [%2] %%asi, %%f4       \n\t"
 145           "ldda [%2+2] %%asi, %%f20    \n\t"
 146           "ld [%0], %%f0               \n\t"
 147           "ld [%0+4], %%f16            \n\t"
 148           "fpmerge %%f3, %%f3, %%f2    \n\t"
 149           "fpmerge %%f19, %%f19, %%f18 \n\t"
 150           "fpmerge %%f5, %%f5, %%f4    \n\t"
 151           "fpmerge %%f21, %%f21, %%f20 \n\t"
 152           YUV2RGB_KERNEL
 153           "fzero %%f0 \n\t"
 154           "fpmerge %%f4, %%f6, %%f8    \n\t"  // r,b,t1
 155           "fpmerge %%f20, %%f22, %%f24 \n\t"  // r,b,t1
 156           "fpmerge %%f0, %%f2, %%f10   \n\t"  // 0,g,t2
 157           "fpmerge %%f0, %%f18, %%f26  \n\t"  // 0,g,t2
 158           "fpmerge %%f10, %%f8, %%f4   \n\t"  // t2,t1,msb
 159           "fpmerge %%f26, %%f24, %%f20 \n\t"  // t2,t1,msb
 160           "fpmerge %%f11, %%f9, %%f6   \n\t"  // t2,t1,lsb
 161           "fpmerge %%f27, %%f25, %%f22 \n\t"  // t2,t1,lsb
 162           "std %%f4, [%3]              \n\t"
 163           "std %%f20, [%3+16]          \n\t"
 164           "std %%f6, [%3+8]            \n\t"
 165           "std %%f22, [%3+24]          \n\t"
 166
 167           "add %0, 8, %0   \n\t"
 168           "add %1, 4, %1   \n\t"
 169           "add %2, 4, %2   \n\t"
 170           "subcc %4, 8, %4 \n\t"
 171           "bne 1b          \n\t"
 172           "add %3, 32, %3  \n\t" //delay slot
 173           : "=r" (out1), "=r" (out2), "=r" (out3), "=r" (out4), "=r" (out5), "=r" (out6)
 174           : "0" (src[0]+(y+srcSliceY)*srcStride[0]), "1" (src[1]+(y+srcSliceY)*srcStride[1]),
 175             "2" (src[2]+(y+srcSliceY)*srcStride[2]), "3" (dst[0]+(y+srcSliceY)*dstStride[0]),
 176             "4" (c->dstW),
 177             "5" (c->sparc_coeffs)
 178       );
 179   }
 180
 181   return srcSliceH;
 182 }
 183
 184 SwsFunc yuv2rgb_init_vis(SwsContext *c) {
 185     c->sparc_coeffs[5]=c->yCoeff;
 186     c->sparc_coeffs[6]=c->vgCoeff;
 187     c->sparc_coeffs[7]=c->vrCoeff;
 188     c->sparc_coeffs[8]=c->ubCoeff;
 189     c->sparc_coeffs[9]=c->ugCoeff;
 190
 191     c->sparc_coeffs[0]=(((int16_t)c->yOffset*(int16_t)c->yCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
 192     c->sparc_coeffs[1]=(((int16_t)c->uOffset*(int16_t)c->ubCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
 193     c->sparc_coeffs[2]=(((int16_t)c->uOffset*(int16_t)c->ugCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
 194     c->sparc_coeffs[3]=(((int16_t)c->vOffset*(int16_t)c->vgCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
 195     c->sparc_coeffs[4]=(((int16_t)c->vOffset*(int16_t)c->vrCoeff>>11) & 0xffff) * 0x0001000100010001ULL;
 196
 197     if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV422P && (c->dstW & 7)==0) {
 198         av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV422P -> RGB32\n");
 199         return vis_422P_ARGB32;
 200     }
 201     else if (c->dstFormat == PIX_FMT_RGB32 && c->srcFormat == PIX_FMT_YUV420P && (c->dstW & 7)==0) {
 202         av_log(c, AV_LOG_INFO, "SPARC VIS accelerated YUV420P -> RGB32\n");
 203         return vis_420P_ARGB32;
 204     }
 205     return NULL;
 206 }