X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.c;h=5c1039b028a801bfe3431a6518c37c623c9b6bc1;hb=3ab770001817e0f52114a9876819f07fcd8ed93a;hp=d3cd6430eb8eea0dbd61f69c945371ebf92843cb;hpb=af818f7a6612e70320f83c7ef0ddc9bfd035d18b;p=ffmpeg diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index d3cd6430eb8..5c1039b028a 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -5,59 +5,54 @@ * * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer * - * This file is part of FFmpeg. + * This file is part of Libav. * - * FFmpeg is free software; you can redistribute it and/or + * Libav is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * FFmpeg is distributed in the hope that it will be useful, + * Libav is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with FFmpeg; if not, write to the Free Software + * License along with Libav; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ /** - * @file libavcodec/dsputil.c + * @file * DSP utils */ +#include "libavutil/imgutils.h" #include "avcodec.h" #include "dsputil.h" #include "simple_idct.h" #include "faandct.h" #include "faanidct.h" #include "mathops.h" -#include "h263.h" -#include "snow.h" #include "mpegvideo.h" #include "config.h" +#include "ac3dec.h" +#include "vorbis.h" +#include "png.h" -/* snow.c */ -void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type, int decomposition_count); - -/* vorbis.c */ -void vorbis_inverse_coupling(float *mag, float *ang, int blocksize); - -/* ac3dec.c */ -void ff_ac3_downmix_c(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len); - -/* lpc.c */ -void ff_lpc_compute_autocorr(const int32_t *data, int len, int lag, double *autoc); +uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, }; +uint32_t ff_squareTbl[512] = {0, }; -/* pngdec.c */ -void ff_add_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp); +#define BIT_DEPTH 9 +#include "dsputil_template.c" +#undef BIT_DEPTH -/* eaidct.c */ -void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block); +#define BIT_DEPTH 10 +#include "dsputil_template.c" +#undef BIT_DEPTH -uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, }; -uint32_t ff_squareTbl[512] = {0, }; +#define BIT_DEPTH 8 +#include "dsputil_template.c" // 0x7f7f7f7f or 0x7f7f7f7f7f7f7f7f or whatever, depending on the cpu's native arithmetic size #define pb_7f (~0UL/255 * 0x7f) @@ -88,7 +83,7 @@ const uint8_t ff_zigzag248_direct[64] = { }; /* not permutated inverse zigzag_direct + 1 for MMX quantizer */ -DECLARE_ALIGNED_16(uint16_t, inv_zigzag_direct16[64]); +DECLARE_ALIGNED(16, uint16_t, inv_zigzag_direct16)[64]; const uint8_t ff_alternate_horizontal_scan[64] = { 0, 1, 2, 3, 8, 9, 16, 17, @@ -112,44 +107,6 @@ const uint8_t ff_alternate_vertical_scan[64] = { 38, 46, 54, 62, 39, 47, 55, 63, }; -/* a*inverse[b]>>32 == a/b for all 0<=a<=16909558 && 2<=b<=256 - * for a>16909558, is an overestimate by less than 1 part in 1<<24 */ -const uint32_t ff_inverse[257]={ - 0, 4294967295U,2147483648U,1431655766, 1073741824, 858993460, 715827883, 613566757, - 536870912, 477218589, 429496730, 390451573, 357913942, 330382100, 306783379, 286331154, - 268435456, 252645136, 238609295, 226050911, 214748365, 204522253, 195225787, 186737709, - 178956971, 171798692, 165191050, 159072863, 153391690, 148102321, 143165577, 138547333, - 134217728, 130150525, 126322568, 122713352, 119304648, 116080198, 113025456, 110127367, - 107374183, 104755300, 102261127, 99882961, 97612894, 95443718, 93368855, 91382283, - 89478486, 87652394, 85899346, 84215046, 82595525, 81037119, 79536432, 78090315, - 76695845, 75350304, 74051161, 72796056, 71582789, 70409300, 69273667, 68174085, - 67108864, 66076420, 65075263, 64103990, 63161284, 62245903, 61356676, 60492498, - 59652324, 58835169, 58040099, 57266231, 56512728, 55778797, 55063684, 54366675, - 53687092, 53024288, 52377650, 51746594, 51130564, 50529028, 49941481, 49367441, - 48806447, 48258060, 47721859, 47197443, 46684428, 46182445, 45691142, 45210183, - 44739243, 44278014, 43826197, 43383509, 42949673, 42524429, 42107523, 41698712, - 41297763, 40904451, 40518560, 40139882, 39768216, 39403370, 39045158, 38693400, - 38347923, 38008561, 37675152, 37347542, 37025581, 36709123, 36398028, 36092163, - 35791395, 35495598, 35204650, 34918434, 34636834, 34359739, 34087043, 33818641, - 33554432, 33294321, 33038210, 32786010, 32537632, 32292988, 32051995, 31814573, - 31580642, 31350127, 31122952, 30899046, 30678338, 30460761, 30246249, 30034737, - 29826162, 29620465, 29417585, 29217465, 29020050, 28825284, 28633116, 28443493, - 28256364, 28071682, 27889399, 27709467, 27531842, 27356480, 27183338, 27012373, - 26843546, 26676816, 26512144, 26349493, 26188825, 26030105, 25873297, 25718368, - 25565282, 25414008, 25264514, 25116768, 24970741, 24826401, 24683721, 24542671, - 24403224, 24265352, 24129030, 23994231, 23860930, 23729102, 23598722, 23469767, - 23342214, 23216040, 23091223, 22967740, 22845571, 22724695, 22605092, 22486740, - 22369622, 22253717, 22139007, 22025474, 21913099, 21801865, 21691755, 21582751, - 21474837, 21367997, 21262215, 21157475, 21053762, 20951060, 20849356, 20748635, - 20648882, 20550083, 20452226, 20355296, 20259280, 20164166, 20069941, 19976593, - 19884108, 19792477, 19701685, 19611723, 19522579, 19434242, 19346700, 19259944, - 19173962, 19088744, 19004281, 18920561, 18837576, 18755316, 18673771, 18592933, - 18512791, 18433337, 18354562, 18276457, 18199014, 18122225, 18046082, 17970575, - 17895698, 17821442, 17747799, 17674763, 17602325, 17530479, 17459217, 17388532, - 17318417, 17248865, 17179870, 17111424, 17043522, 16976156, 16909321, 16843010, - 16777216 -}; - /* Input permutation for the simple_idct_mmx */ static const uint8_t simple_mmx_permutation[64]={ 0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D, @@ -188,6 +145,41 @@ void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_s } } +void ff_init_scantable_permutation(uint8_t *idct_permutation, + int idct_permutation_type) +{ + int i; + + switch(idct_permutation_type){ + case FF_NO_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= i; + break; + case FF_LIBMPEG2_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2); + break; + case FF_SIMPLE_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= simple_mmx_permutation[i]; + break; + case FF_TRANSPOSE_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= ((i&7)<<3) | (i>>3); + break; + case FF_PARTTRANS_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= (i&0x24) | ((i&3)<<3) | ((i>>3)&3); + break; + case FF_SSE2_IDCT_PERM: + for(i=0; i<64; i++) + idct_permutation[i]= (i&0x38) | idct_sse2_row_perm[i&7]; + break; + default: + av_log(NULL, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n"); + } +} + static int pix_sum_c(uint8_t * pix, int line_size) { int s, i, j; @@ -228,7 +220,7 @@ static int pix_norm1_c(uint8_t * pix, int line_size) s += sq[pix[6]]; s += sq[pix[7]]; #else -#if LONG_MAX > 2147483647 +#if HAVE_FAST_64BIT register uint64_t x=*(uint64_t*)pix; s += sq[x&0xff]; s += sq[(x>>8)&0xff]; @@ -262,20 +254,26 @@ static void bswap_buf(uint32_t *dst, const uint32_t *src, int w){ int i; for(i=0; i+8<=w; i+=8){ - dst[i+0]= bswap_32(src[i+0]); - dst[i+1]= bswap_32(src[i+1]); - dst[i+2]= bswap_32(src[i+2]); - dst[i+3]= bswap_32(src[i+3]); - dst[i+4]= bswap_32(src[i+4]); - dst[i+5]= bswap_32(src[i+5]); - dst[i+6]= bswap_32(src[i+6]); - dst[i+7]= bswap_32(src[i+7]); + dst[i+0]= av_bswap32(src[i+0]); + dst[i+1]= av_bswap32(src[i+1]); + dst[i+2]= av_bswap32(src[i+2]); + dst[i+3]= av_bswap32(src[i+3]); + dst[i+4]= av_bswap32(src[i+4]); + dst[i+5]= av_bswap32(src[i+5]); + dst[i+6]= av_bswap32(src[i+6]); + dst[i+7]= av_bswap32(src[i+7]); } for(;i>(dec_count-level); - int sx= (ori&1) ? size : 0; - int stride= 32<<(dec_count-level); - int sy= (ori&2) ? stride>>1 : 0; - - for(i=0; i=0); - return s>>9; -} - -static int w53_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 8, h, 1); -} - -static int w97_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 8, h, 0); -} - -static int w53_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 16, h, 1); -} - -static int w97_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 16, h, 0); -} - -int w53_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 32, h, 1); -} - -int w97_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){ - return w_c(v, pix1, pix2, line_size, 32, h, 0); -} -#endif - -/* draw the edges of width 'w' of an image of size width, height */ -//FIXME check that this is ok for mpeg4 interlaced -static void draw_edges_c(uint8_t *buf, int wrap, int width, int height, int w) -{ - uint8_t *ptr, *last_line; - int i; - - last_line = buf + (height - 1) * wrap; - for(i=0;i= h){ - src+= (h-1-src_y)*linesize; - src_y=h-1; - }else if(src_y<=-block_h){ - src+= (1-block_h-src_y)*linesize; - src_y=1-block_h; - } - if(src_x>= w){ - src+= (w-1-src_x); - src_x=w-1; - }else if(src_x<=-block_w){ - src+= (1-block_w-src_x); - src_x=1-block_w; - } - - start_y= FFMAX(0, -src_y); - start_x= FFMAX(0, -src_x); - end_y= FFMIN(block_h, h-src_y); - end_x= FFMIN(block_w, w-src_x); - - // copy existing part - for(y=start_y; y>1));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static void OPNAME ## _pixels_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - for(i=0; i>1));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static void OPNAME ## _no_rnd_pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - for(i=0; i>1));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static void OPNAME ## _pixels_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - for(i=0; i>1));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static void OPNAME ## _pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - const uint64_t a= AV_RN64(pixels );\ - const uint64_t b= AV_RN64(pixels+1);\ - uint64_t l0= (a&0x0303030303030303ULL)\ - + (b&0x0303030303030303ULL)\ - + 0x0202020202020202ULL;\ - uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - uint64_t l1,h1;\ -\ - pixels+=line_size;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ - pixels+=line_size;\ - block +=line_size;\ - a= AV_RN64(pixels );\ - b= AV_RN64(pixels+1);\ - l0= (a&0x0303030303030303ULL)\ - + (b&0x0303030303030303ULL)\ - + 0x0202020202020202ULL;\ - h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static void OPNAME ## _no_rnd_pixels_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - const uint64_t a= AV_RN64(pixels );\ - const uint64_t b= AV_RN64(pixels+1);\ - uint64_t l0= (a&0x0303030303030303ULL)\ - + (b&0x0303030303030303ULL)\ - + 0x0101010101010101ULL;\ - uint64_t h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - uint64_t l1,h1;\ -\ - pixels+=line_size;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ - pixels+=line_size;\ - block +=line_size;\ - a= AV_RN64(pixels );\ - b= AV_RN64(pixels+1);\ - l0= (a&0x0303030303030303ULL)\ - + (b&0x0303030303030303ULL)\ - + 0x0101010101010101ULL;\ - h0= ((a&0xFCFCFCFCFCFCFCFCULL)>>2)\ - + ((b&0xFCFCFCFCFCFCFCFCULL)>>2);\ - OP(*((uint64_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0F0F0F0F0FULL));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -CALL_2X_PIXELS(OPNAME ## _pixels16_c , OPNAME ## _pixels_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels_x2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels_y2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels_xy2_c, 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels_x2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels_y2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels_xy2_c, 8) - -#define op_avg(a, b) a = ( ((a)|(b)) - ((((a)^(b))&0xFEFEFEFEFEFEFEFEULL)>>1) ) -#else // 64 bit variant - -#define PIXOP2(OPNAME, OP) \ -static void OPNAME ## _pixels2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ - int i;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - l1= (c&0x03030303UL)\ - + (d&0x03030303UL);\ - h1= ((c&0xFCFCFCFCUL)>>2)\ - + ((d&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - a= AV_RN32(&src1[i*src_stride1+4]);\ - b= AV_RN32(&src2[i*src_stride2+4]);\ - c= AV_RN32(&src3[i*src_stride3+4]);\ - d= AV_RN32(&src4[i*src_stride4+4]);\ - l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x02020202UL;\ - h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - l1= (c&0x03030303UL)\ - + (d&0x03030303UL);\ - h1= ((c&0xFCFCFCFCUL)>>2)\ - + ((d&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - }\ -}\ -\ -static inline void OPNAME ## _pixels4_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ - OPNAME ## _pixels4_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\ -}\ -\ -static inline void OPNAME ## _pixels4_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ - OPNAME ## _pixels4_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\ -}\ -\ -static inline void OPNAME ## _pixels2_x2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ - OPNAME ## _pixels2_l2(block, pixels, pixels+1, line_size, line_size, line_size, h);\ -}\ -\ -static inline void OPNAME ## _pixels2_y2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h){\ - OPNAME ## _pixels2_l2(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\ -}\ -\ -static inline void OPNAME ## _no_rnd_pixels8_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\ - int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\ - int i;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - l1= (c&0x03030303UL)\ - + (d&0x03030303UL);\ - h1= ((c&0xFCFCFCFCUL)>>2)\ - + ((d&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)&dst[i*dst_stride]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - a= AV_RN32(&src1[i*src_stride1+4]);\ - b= AV_RN32(&src2[i*src_stride2+4]);\ - c= AV_RN32(&src3[i*src_stride3+4]);\ - d= AV_RN32(&src4[i*src_stride4+4]);\ - l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x01010101UL;\ - h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - l1= (c&0x03030303UL)\ - + (d&0x03030303UL);\ - h1= ((c&0xFCFCFCFCUL)>>2)\ - + ((d&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)&dst[i*dst_stride+4]), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - }\ -}\ -static inline void OPNAME ## _pixels16_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\ - int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\ - OPNAME ## _pixels8_l4(dst , src1 , src2 , src3 , src4 , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\ - OPNAME ## _pixels8_l4(dst+8, src1+8, src2+8, src3+8, src4+8, dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\ -}\ -static inline void OPNAME ## _no_rnd_pixels16_l4(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,\ - int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\ - OPNAME ## _no_rnd_pixels8_l4(dst , src1 , src2 , src3 , src4 , dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\ - OPNAME ## _no_rnd_pixels8_l4(dst+8, src1+8, src2+8, src3+8, src4+8, dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\ -}\ -\ -static inline void OPNAME ## _pixels2_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i, a0, b0, a1, b1;\ - a0= pixels[0];\ - b0= pixels[1] + 2;\ - a0 += b0;\ - b0 += pixels[2];\ -\ - pixels+=line_size;\ - for(i=0; i>2; /* FIXME non put */\ - block[1]= (b1+b0)>>2;\ -\ - pixels+=line_size;\ - block +=line_size;\ -\ - a0= pixels[0];\ - b0= pixels[1] + 2;\ - a0 += b0;\ - b0 += pixels[2];\ -\ - block[0]= (a1+a0)>>2;\ - block[1]= (b1+b0)>>2;\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static inline void OPNAME ## _pixels4_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int i;\ - const uint32_t a= AV_RN32(pixels );\ - const uint32_t b= AV_RN32(pixels+1);\ - uint32_t l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x02020202UL;\ - uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - uint32_t l1,h1;\ -\ - pixels+=line_size;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - a= AV_RN32(pixels );\ - b= AV_RN32(pixels+1);\ - l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x02020202UL;\ - h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - }\ -}\ -\ -static inline void OPNAME ## _pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int j;\ - for(j=0; j<2; j++){\ - int i;\ - const uint32_t a= AV_RN32(pixels );\ - const uint32_t b= AV_RN32(pixels+1);\ - uint32_t l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x02020202UL;\ - uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - uint32_t l1,h1;\ -\ - pixels+=line_size;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - a= AV_RN32(pixels );\ - b= AV_RN32(pixels+1);\ - l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x02020202UL;\ - h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - }\ - pixels+=4-line_size*(h+1);\ - block +=4-line_size*h;\ - }\ -}\ -\ -static inline void OPNAME ## _no_rnd_pixels8_xy2_c(uint8_t *block, const uint8_t *pixels, int line_size, int h)\ -{\ - int j;\ - for(j=0; j<2; j++){\ - int i;\ - const uint32_t a= AV_RN32(pixels );\ - const uint32_t b= AV_RN32(pixels+1);\ - uint32_t l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x01010101UL;\ - uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - uint32_t l1,h1;\ -\ - pixels+=line_size;\ - for(i=0; i>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - a= AV_RN32(pixels );\ - b= AV_RN32(pixels+1);\ - l0= (a&0x03030303UL)\ - + (b&0x03030303UL)\ - + 0x01010101UL;\ - h0= ((a&0xFCFCFCFCUL)>>2)\ - + ((b&0xFCFCFCFCUL)>>2);\ - OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\ - pixels+=line_size;\ - block +=line_size;\ - }\ - pixels+=4-line_size*(h+1);\ - block +=4-line_size*h;\ - }\ -}\ -\ -CALL_2X_PIXELS(OPNAME ## _pixels16_c , OPNAME ## _pixels8_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_x2_c , OPNAME ## _pixels8_x2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_y2_c , OPNAME ## _pixels8_y2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_c, OPNAME ## _pixels8_xy2_c, 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_c , OPNAME ## _pixels8_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_c , OPNAME ## _no_rnd_pixels8_x2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_c , OPNAME ## _no_rnd_pixels8_y2_c , 8)\ -CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_c, OPNAME ## _no_rnd_pixels8_xy2_c, 8)\ - -#define op_avg(a, b) a = rnd_avg32(a, b) -#endif -#define op_put(a, b) a = b +static void fill_block16_c(uint8_t *block, uint8_t value, int line_size, int h) +{ + int i; -PIXOP2(avg, op_avg) -PIXOP2(put, op_put) -#undef op_avg -#undef op_put + for (i = 0; i < h; i++) { + memset(block, value, 16); + block += line_size; + } +} -#define avg2(a,b) ((a+b+1)>>1) -#define avg4(a,b,c,d) ((a+b+c+d+2)>>2) +static void fill_block8_c(uint8_t *block, uint8_t value, int line_size, int h) +{ + int i; -static void put_no_rnd_pixels16_l2_c(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){ - put_no_rnd_pixels16_l2(dst, a, b, stride, stride, stride, h); + for (i = 0; i < h; i++) { + memset(block, value, 8); + block += line_size; + } } -static void put_no_rnd_pixels8_l2_c(uint8_t *dst, const uint8_t *a, const uint8_t *b, int stride, int h){ - put_no_rnd_pixels8_l2(dst, a, b, stride, stride, stride, h); -} +#define avg2(a,b) ((a+b+1)>>1) +#define avg4(a,b,c,d) ((a+b+c+d+2)>>2) static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y16, int rounder) { @@ -1361,10 +607,10 @@ void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, static inline void put_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ switch(width){ - case 2: put_pixels2_c (dst, src, stride, height); break; - case 4: put_pixels4_c (dst, src, stride, height); break; - case 8: put_pixels8_c (dst, src, stride, height); break; - case 16:put_pixels16_c(dst, src, stride, height); break; + case 2: put_pixels2_8_c (dst, src, stride, height); break; + case 4: put_pixels4_8_c (dst, src, stride, height); break; + case 8: put_pixels8_8_c (dst, src, stride, height); break; + case 16:put_pixels16_8_c(dst, src, stride, height); break; } } @@ -1458,10 +704,10 @@ static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int static inline void avg_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ switch(width){ - case 2: avg_pixels2_c (dst, src, stride, height); break; - case 4: avg_pixels4_c (dst, src, stride, height); break; - case 8: avg_pixels8_c (dst, src, stride, height); break; - case 16:avg_pixels16_c(dst, src, stride, height); break; + case 2: avg_pixels2_8_c (dst, src, stride, height); break; + case 4: avg_pixels4_8_c (dst, src, stride, height); break; + case 8: avg_pixels8_8_c (dst, src, stride, height); break; + case 16:avg_pixels16_8_c(dst, src, stride, height); break; } } @@ -1552,184 +798,6 @@ static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int dst += stride; } } -#if 0 -#define TPEL_WIDTH(width)\ -static void put_tpel_pixels ## width ## _mc00_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc00_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc10_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc10_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc20_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc20_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc01_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc01_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc11_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc11_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc21_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc21_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc02_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc02_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc12_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc12_c(dst, src, stride, width, height);}\ -static void put_tpel_pixels ## width ## _mc22_c(uint8_t *dst, const uint8_t *src, int stride, int height){\ - void put_tpel_pixels_mc22_c(dst, src, stride, width, height);} -#endif - -#define H264_CHROMA_MC(OPNAME, OP)\ -static void OPNAME ## h264_chroma_mc2_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\ - const int A=(8-x)*(8-y);\ - const int B=( x)*(8-y);\ - const int C=(8-x)*( y);\ - const int D=( x)*( y);\ - int i;\ - \ - assert(x<8 && y<8 && x>=0 && y>=0);\ -\ - if(D){\ - for(i=0; i=0 && y>=0);\ -\ - if(D){\ - for(i=0; i=0 && y>=0);\ -\ - if(D){\ - for(i=0; i>6)+1)>>1) -#define op_put(a, b) a = (((b) + 32)>>6) - -H264_CHROMA_MC(put_ , op_put) -H264_CHROMA_MC(avg_ , op_avg) -#undef op_avg -#undef op_put - -static void put_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){ - const int A=(8-x)*(8-y); - const int B=( x)*(8-y); - const int C=(8-x)*( y); - const int D=( x)*( y); - int i; - - assert(x<8 && y<8 && x>=0 && y>=0); - - for(i=0; i> 6; - dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6; - dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6; - dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6; - dst[4] = (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6; - dst[5] = (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6; - dst[6] = (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6; - dst[7] = (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6; - dst+= stride; - src+= stride; - } -} - -static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){ - const int A=(8-x)*(8-y); - const int B=( x)*(8-y); - const int C=(8-x)*( y); - const int D=( x)*( y); - int i; - - assert(x<8 && y<8 && x>=0 && y>=0); - - for(i=0; i> 6)); - dst[1] = avg2(dst[1], ((A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6)); - dst[2] = avg2(dst[2], ((A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6)); - dst[3] = avg2(dst[3], ((A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6)); - dst[4] = avg2(dst[4], ((A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6)); - dst[5] = avg2(dst[5], ((A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6)); - dst[6] = avg2(dst[6], ((A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6)); - dst[7] = avg2(dst[7], ((A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6)); - dst+= stride; - src+= stride; - } -} #define QPEL_MC(r, OPNAME, RND, OP) \ static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\ @@ -1849,14 +917,10 @@ static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dst }\ }\ \ -static void OPNAME ## qpel8_mc00_c (uint8_t *dst, uint8_t *src, int stride){\ - OPNAME ## pixels8_c(dst, src, stride, 8);\ -}\ -\ static void OPNAME ## qpel8_mc10_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[64];\ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\ - OPNAME ## pixels8_l2(dst, src, half, stride, stride, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, src, half, stride, stride, 8, 8);\ }\ \ static void OPNAME ## qpel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -1866,7 +930,7 @@ static void OPNAME ## qpel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){\ static void OPNAME ## qpel8_mc30_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[64];\ put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\ - OPNAME ## pixels8_l2(dst, src+1, half, stride, stride, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, src+1, half, stride, stride, 8, 8);\ }\ \ static void OPNAME ## qpel8_mc01_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -1874,7 +938,7 @@ static void OPNAME ## qpel8_mc01_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\ - OPNAME ## pixels8_l2(dst, full, half, stride, 16, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, full, half, stride, 16, 8, 8);\ }\ \ static void OPNAME ## qpel8_mc02_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -1888,7 +952,7 @@ static void OPNAME ## qpel8_mc03_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\ - OPNAME ## pixels8_l2(dst, full+16, half, stride, 16, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, full+16, half, stride, 16, 8, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1899,7 +963,7 @@ void ff_ ## OPNAME ## qpel8_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l4(dst, full, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ + OPNAME ## pixels8_l4_8(dst, full, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc11_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1907,9 +971,9 @@ static void OPNAME ## qpel8_mc11_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1920,7 +984,7 @@ void ff_ ## OPNAME ## qpel8_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l4(dst, full+1, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ + OPNAME ## pixels8_l4_8(dst, full+1, halfH, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc31_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1928,9 +992,9 @@ static void OPNAME ## qpel8_mc31_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full+1, 8, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1941,7 +1005,7 @@ void ff_ ## OPNAME ## qpel8_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l4(dst, full+16, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ + OPNAME ## pixels8_l4_8(dst, full+16, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc13_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1949,9 +1013,9 @@ static void OPNAME ## qpel8_mc13_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH+8, halfHV, stride, 8, 8, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1962,7 +1026,7 @@ void ff_ ## OPNAME ## qpel8_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full , 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l4(dst, full+17, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ + OPNAME ## pixels8_l4_8(dst, full+17, halfH+8, halfV, halfHV, stride, 16, 8, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1970,23 +1034,23 @@ static void OPNAME ## qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[64];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full+1, 8, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH+8, halfHV, stride, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc21_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfH[72];\ uint8_t halfHV[64];\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH, halfHV, stride, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc23_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfH[72];\ uint8_t halfHV[64];\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfH+8, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfH+8, halfHV, stride, 8, 8, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc12_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ @@ -1997,14 +1061,14 @@ void ff_ ## OPNAME ## qpel8_mc12_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc12_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ uint8_t halfH[72];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full, 8, 8, 16, 9);\ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\ }\ void ff_ ## OPNAME ## qpel8_mc32_old_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -2016,14 +1080,14 @@ void ff_ ## OPNAME ## qpel8_mc32_old_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfV, full+1, 8, 16);\ put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\ - OPNAME ## pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8);\ + OPNAME ## pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8);\ }\ static void OPNAME ## qpel8_mc32_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[16*9];\ uint8_t halfH[72];\ copy_block9(full, src, 16, stride, 9);\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\ - put ## RND ## pixels8_l2(halfH, halfH, full+1, 8, 8, 16, 9);\ + put ## RND ## pixels8_l2_8(halfH, halfH, full+1, 8, 8, 16, 9);\ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\ }\ static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -2031,14 +1095,11 @@ static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){\ put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\ OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\ }\ -static void OPNAME ## qpel16_mc00_c (uint8_t *dst, uint8_t *src, int stride){\ - OPNAME ## pixels16_c(dst, src, stride, 16);\ -}\ \ static void OPNAME ## qpel16_mc10_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[256];\ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\ - OPNAME ## pixels16_l2(dst, src, half, stride, stride, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, src, half, stride, stride, 16, 16);\ }\ \ static void OPNAME ## qpel16_mc20_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -2048,7 +1109,7 @@ static void OPNAME ## qpel16_mc20_c(uint8_t *dst, uint8_t *src, int stride){\ static void OPNAME ## qpel16_mc30_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[256];\ put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\ - OPNAME ## pixels16_l2(dst, src+1, half, stride, stride, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, src+1, half, stride, stride, 16, 16);\ }\ \ static void OPNAME ## qpel16_mc01_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -2056,7 +1117,7 @@ static void OPNAME ## qpel16_mc01_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[256];\ copy_block17(full, src, 24, stride, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\ - OPNAME ## pixels16_l2(dst, full, half, stride, 24, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, full, half, stride, 24, 16, 16);\ }\ \ static void OPNAME ## qpel16_mc02_c(uint8_t *dst, uint8_t *src, int stride){\ @@ -2070,7 +1131,7 @@ static void OPNAME ## qpel16_mc03_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t half[256];\ copy_block17(full, src, 24, stride, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\ - OPNAME ## pixels16_l2(dst, full+24, half, stride, 24, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, full+24, half, stride, 24, 16, 16);\ }\ void ff_ ## OPNAME ## qpel16_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2081,7 +1142,7 @@ void ff_ ## OPNAME ## qpel16_mc11_old_c(uint8_t *dst, uint8_t *src, int stride){ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l4(dst, full, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ + OPNAME ## pixels16_l4_8(dst, full, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ }\ static void OPNAME ## qpel16_mc11_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2089,9 +1150,9 @@ static void OPNAME ## qpel16_mc11_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[256];\ copy_block17(full, src, 24, stride, 17);\ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\ + put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16);\ }\ void ff_ ## OPNAME ## qpel16_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2102,7 +1163,7 @@ void ff_ ## OPNAME ## qpel16_mc31_old_c(uint8_t *dst, uint8_t *src, int stride){ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l4(dst, full+1, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ + OPNAME ## pixels16_l4_8(dst, full+1, halfH, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ }\ static void OPNAME ## qpel16_mc31_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2110,9 +1171,9 @@ static void OPNAME ## qpel16_mc31_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[256];\ copy_block17(full, src, 24, stride, 17);\ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\ + put ## RND ## pixels16_l2_8(halfH, halfH, full+1, 16, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, halfH, halfHV, stride, 16, 16, 16);\ }\ void ff_ ## OPNAME ## qpel16_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2123,7 +1184,7 @@ void ff_ ## OPNAME ## qpel16_mc13_old_c(uint8_t *dst, uint8_t *src, int stride){ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l4(dst, full+24, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ + OPNAME ## pixels16_l4_8(dst, full+24, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ }\ static void OPNAME ## qpel16_mc13_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2131,9 +1192,9 @@ static void OPNAME ## qpel16_mc13_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t halfHV[256];\ copy_block17(full, src, 24, stride, 17);\ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\ + put ## RND ## pixels16_l2_8(halfH, halfH, full, 16, 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\ + OPNAME ## pixels16_l2_8(dst, halfH+16, halfHV, stride, 16, 16, 16);\ }\ void ff_ ## OPNAME ## qpel16_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){\ uint8_t full[24*17];\ @@ -2144,583 +1205,96 @@ void ff_ ## OPNAME ## qpel16_mc33_old_c(uint8_t *dst, uint8_t *src, int stride){ put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full , 16, 24, 17);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\ put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l4(dst, full+25, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ + OPNAME ## pixels16_l4_8(dst, full+25, halfH+16, halfV, halfHV, stride, 24, 16, 16, 16, 16);\ }\ static void OPNAME ## qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[24*17];\ - uint8_t halfH[272];\ - uint8_t halfHV[256];\ - copy_block17(full, src, 24, stride, 17);\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\ -}\ -static void OPNAME ## qpel16_mc21_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t halfH[272];\ - uint8_t halfHV[256];\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH, halfHV, stride, 16, 16, 16);\ -}\ -static void OPNAME ## qpel16_mc23_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t halfH[272];\ - uint8_t halfHV[256];\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfH+16, halfHV, stride, 16, 16, 16);\ -}\ -void ff_ ## OPNAME ## qpel16_mc12_old_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[24*17];\ - uint8_t halfH[272];\ - uint8_t halfV[256];\ - uint8_t halfHV[256];\ - copy_block17(full, src, 24, stride, 17);\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full, 16, 24);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfV, halfHV, stride, 16, 16, 16);\ -}\ -static void OPNAME ## qpel16_mc12_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[24*17];\ - uint8_t halfH[272];\ - copy_block17(full, src, 24, stride, 17);\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full, 16, 16, 24, 17);\ - OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\ -}\ -void ff_ ## OPNAME ## qpel16_mc32_old_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[24*17];\ - uint8_t halfH[272];\ - uint8_t halfV[256];\ - uint8_t halfHV[256];\ - copy_block17(full, src, 24, stride, 17);\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfV, full+1, 16, 24);\ - put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\ - OPNAME ## pixels16_l2(dst, halfV, halfHV, stride, 16, 16, 16);\ -}\ -static void OPNAME ## qpel16_mc32_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t full[24*17];\ - uint8_t halfH[272];\ - copy_block17(full, src, 24, stride, 17);\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\ - put ## RND ## pixels16_l2(halfH, halfH, full+1, 16, 16, 24, 17);\ - OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\ -}\ -static void OPNAME ## qpel16_mc22_c(uint8_t *dst, uint8_t *src, int stride){\ - uint8_t halfH[272];\ - put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\ - OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\ -} - -#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1) -#define op_avg_no_rnd(a, b) a = (((a)+cm[((b) + 15)>>5])>>1) -#define op_put(a, b) a = cm[((b) + 16)>>5] -#define op_put_no_rnd(a, b) a = cm[((b) + 15)>>5] - -QPEL_MC(0, put_ , _ , op_put) -QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd) -QPEL_MC(0, avg_ , _ , op_avg) -//QPEL_MC(1, avg_no_rnd , _ , op_avg) -#undef op_avg -#undef op_avg_no_rnd -#undef op_put -#undef op_put_no_rnd - -#if 1 -#define H264_LOWPASS(OPNAME, OP, OP2) \ -static av_unused void OPNAME ## h264_qpel2_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\ - const int h=2;\ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\ - int i;\ - for(i=0; i>5]+1)>>1) -//#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7) -#define op_put(a, b) a = cm[((b) + 16)>>5] -#define op2_avg(a, b) a = (((a)+cm[((b) + 512)>>10]+1)>>1) -#define op2_put(a, b) a = cm[((b) + 512)>>10] - -H264_LOWPASS(put_ , op_put, op2_put) -H264_LOWPASS(avg_ , op_avg, op2_avg) -H264_MC(put_, 2) -H264_MC(put_, 4) -H264_MC(put_, 8) -H264_MC(put_, 16) -H264_MC(avg_, 4) -H264_MC(avg_, 8) -H264_MC(avg_, 16) +#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1) +#define op_avg_no_rnd(a, b) a = (((a)+cm[((b) + 15)>>5])>>1) +#define op_put(a, b) a = cm[((b) + 16)>>5] +#define op_put_no_rnd(a, b) a = cm[((b) + 15)>>5] +QPEL_MC(0, put_ , _ , op_put) +QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd) +QPEL_MC(0, avg_ , _ , op_avg) +//QPEL_MC(1, avg_no_rnd , _ , op_avg) #undef op_avg +#undef op_avg_no_rnd #undef op_put -#undef op2_avg -#undef op2_put -#endif - -#define op_scale1(x) block[x] = av_clip_uint8( (block[x]*weight + offset) >> log2_denom ) -#define op_scale2(x) dst[x] = av_clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1)) -#define H264_WEIGHT(W,H) \ -static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \ - int y; \ - offset <<= log2_denom; \ - if(log2_denom) offset += 1<<(log2_denom-1); \ - for(y=0; y> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] ); - tc++; - } - if( FFABS( q2 - q0 ) < beta ) { - pix[ xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] ); - tc++; - } - - i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc ); - pix[-xstride] = av_clip_uint8( p0 + i_delta ); /* p0' */ - pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */ - } - pix += ystride; - } - } -} -static void h264_v_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) -{ - h264_loop_filter_luma_c(pix, stride, 1, alpha, beta, tc0); -} -static void h264_h_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) -{ - h264_loop_filter_luma_c(pix, 1, stride, alpha, beta, tc0); -} - -static inline void h264_loop_filter_luma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta) -{ - int d; - for( d = 0; d < 16; d++ ) { - const int p2 = pix[-3*xstride]; - const int p1 = pix[-2*xstride]; - const int p0 = pix[-1*xstride]; - - const int q0 = pix[ 0*xstride]; - const int q1 = pix[ 1*xstride]; - const int q2 = pix[ 2*xstride]; - - if( FFABS( p0 - q0 ) < alpha && - FFABS( p1 - p0 ) < beta && - FFABS( q1 - q0 ) < beta ) { - - if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){ - if( FFABS( p2 - p0 ) < beta) - { - const int p3 = pix[-4*xstride]; - /* p0', p1', p2' */ - pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3; - pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2; - pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3; - } else { - /* p0' */ - pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; - } - if( FFABS( q2 - q0 ) < beta) - { - const int q3 = pix[3*xstride]; - /* q0', q1', q2' */ - pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3; - pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2; - pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3; - } else { - /* q0' */ - pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2; - } - }else{ - /* p0', q0' */ - pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; - pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2; - } - } - pix += ystride; - } -} -static void h264_v_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta) -{ - h264_loop_filter_luma_intra_c(pix, stride, 1, alpha, beta); -} -static void h264_h_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta) -{ - h264_loop_filter_luma_intra_c(pix, 1, stride, alpha, beta); -} - -static inline void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0) -{ - int i, d; - for( i = 0; i < 4; i++ ) { - const int tc = tc0[i]; - if( tc <= 0 ) { - pix += 2*ystride; - continue; - } - for( d = 0; d < 2; d++ ) { - const int p0 = pix[-1*xstride]; - const int p1 = pix[-2*xstride]; - const int q0 = pix[0]; - const int q1 = pix[1*xstride]; - - if( FFABS( p0 - q0 ) < alpha && - FFABS( p1 - p0 ) < beta && - FFABS( q1 - q0 ) < beta ) { - - int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc ); - - pix[-xstride] = av_clip_uint8( p0 + delta ); /* p0' */ - pix[0] = av_clip_uint8( q0 - delta ); /* q0' */ - } - pix += ystride; - } - } -} -static void h264_v_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) -{ - h264_loop_filter_chroma_c(pix, stride, 1, alpha, beta, tc0); -} -static void h264_h_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) -{ - h264_loop_filter_chroma_c(pix, 1, stride, alpha, beta, tc0); -} - -static inline void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta) -{ - int d; - for( d = 0; d < 8; d++ ) { - const int p0 = pix[-1*xstride]; - const int p1 = pix[-2*xstride]; - const int q0 = pix[0]; - const int q1 = pix[1*xstride]; - - if( FFABS( p0 - q0 ) < alpha && - FFABS( p1 - p0 ) < beta && - FFABS( q1 - q0 ) < beta ) { - - pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; /* p0' */ - pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2; /* q0' */ - } - pix += ystride; - } -} -static void h264_v_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta) -{ - h264_loop_filter_chroma_intra_c(pix, stride, 1, alpha, beta); -} -static void h264_h_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta) -{ - h264_loop_filter_chroma_intra_c(pix, 1, stride, alpha, beta); -} - static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) { int s, i; @@ -3423,7 +1779,7 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){ } /** - * permutes an 8x8 block. + * Permute an 8x8 block. * @param block the block which will be permuted according to the given permutation vector * @param permutation the permutation vector * @param last the last non zero coefficient in scantable order, used to speed the permutation up @@ -3501,7 +1857,7 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){ case FF_CMP_NSSE: cmp[i]= c->nsse[i]; break; -#if CONFIG_SNOW_ENCODER +#if CONFIG_DWT case FF_CMP_W53: cmp[i]= c->w53[i]; break; @@ -3515,19 +1871,6 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){ } } -static void clear_block_c(DCTELEM *block) -{ - memset(block, 0, sizeof(DCTELEM)*64); -} - -/** - * memset(blocks, 0, sizeof(DCTELEM)*6*64) - */ -static void clear_blocks_c(DCTELEM *blocks) -{ - memset(blocks, 0, sizeof(DCTELEM)*6*64); -} - static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){ long i; for(i=0; i<=w-sizeof(long); i+=sizeof(long)){ @@ -3633,35 +1976,42 @@ static int add_hfyu_left_prediction_c(uint8_t *dst, const uint8_t *src, int w, i #define B 3 #define G 2 #define R 1 +#define A 0 #else #define B 0 #define G 1 #define R 2 +#define A 3 #endif -static void add_hfyu_left_prediction_bgr32_c(uint8_t *dst, const uint8_t *src, int w, int *red, int *green, int *blue){ +static void add_hfyu_left_prediction_bgr32_c(uint8_t *dst, const uint8_t *src, int w, int *red, int *green, int *blue, int *alpha){ int i; - int r,g,b; + int r,g,b,a; r= *red; g= *green; b= *blue; + a= *alpha; for(i=0; imaxi){ - maxi=sum; - printf("MAX:%d\n", maxi); -} -#endif return sum; } @@ -3780,8 +2123,7 @@ static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_ static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){ MpegEncContext * const s= (MpegEncContext *)c; - DECLARE_ALIGNED_16(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]); - DCTELEM * const temp= (DCTELEM*)aligned_temp; + LOCAL_ALIGNED_16(DCTELEM, temp, [64]); assert(h==8); @@ -3845,8 +2187,7 @@ static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){ MpegEncContext * const s= (MpegEncContext *)c; - DECLARE_ALIGNED_16(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]); - DCTELEM * const temp= (DCTELEM*)aligned_temp; + LOCAL_ALIGNED_16(DCTELEM, temp, [64]); int sum=0, i; assert(h==8); @@ -3862,9 +2203,8 @@ static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2 static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){ MpegEncContext * const s= (MpegEncContext *)c; - DECLARE_ALIGNED_16(uint64_t, aligned_temp[sizeof(DCTELEM)*64*2/8]); - DCTELEM * const temp= (DCTELEM*)aligned_temp; - DCTELEM * const bak = ((DCTELEM*)aligned_temp)+64; + LOCAL_ALIGNED_16(DCTELEM, temp, [64*2]); + DCTELEM * const bak = temp+64; int sum=0, i; assert(h==8); @@ -3876,7 +2216,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i); s->dct_unquantize_inter(s, temp, 0, s->qscale); - ff_simple_idct(temp); //FIXME + ff_simple_idct_8(temp); //FIXME for(i=0; i<64; i++) sum+= (temp[i]-bak[i])*(temp[i]-bak[i]); @@ -3887,12 +2227,9 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){ MpegEncContext * const s= (MpegEncContext *)c; const uint8_t *scantable= s->intra_scantable.permutated; - DECLARE_ALIGNED_16(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]); - DECLARE_ALIGNED_16(uint64_t, aligned_src1[8]); - DECLARE_ALIGNED_16(uint64_t, aligned_src2[8]); - DCTELEM * const temp= (DCTELEM*)aligned_temp; - uint8_t * const lsrc1 = (uint8_t*)aligned_src1; - uint8_t * const lsrc2 = (uint8_t*)aligned_src2; + LOCAL_ALIGNED_16(DCTELEM, temp, [64]); + LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]); + LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]); int i, last, run, bits, level, distortion, start_i; const int esc_length= s->ac_esc_length; uint8_t * length; @@ -3966,8 +2303,7 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){ MpegEncContext * const s= (MpegEncContext *)c; const uint8_t *scantable= s->intra_scantable.permutated; - DECLARE_ALIGNED_16(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]); - DCTELEM * const temp= (DCTELEM*)aligned_temp; + LOCAL_ALIGNED_16(DCTELEM, temp, [64]); int i, last, run, bits, level, start_i; const int esc_length= s->ac_esc_length; uint8_t * length; @@ -4110,10 +2446,10 @@ WRAPPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c) WRAPPER8_16_SQ(rd8x8_c, rd16_c) WRAPPER8_16_SQ(bit8x8_c, bit16_c) -static void vector_fmul_c(float *dst, const float *src, int len){ +static void vector_fmul_c(float *dst, const float *src0, const float *src1, int len){ int i; for(i=0; i mini) return mini; - else if((a^(1<<31)) > maxisign) return maxi; + else if((a^(1U<<31)) > maxisign) return maxi; else return a; } @@ -4237,7 +2545,7 @@ static void vector_clipf_c_opposite_sign(float *dst, const float *src, float *mi int i; uint32_t mini = *(uint32_t*)min; uint32_t maxi = *(uint32_t*)max; - uint32_t maxisign = maxi ^ (1<<31); + uint32_t maxisign = maxi ^ (1U<<31); uint32_t *dsti = (uint32_t*)dst; const uint32_t *srci = (const uint32_t*)src; for(i=0; i>31; - // is this faster on some gcc/cpu combinations? -// if(tmp > 0x43c0ffff) tmp = 0xFFFF; -// else tmp = 0; - } - return tmp - 0x8000; -} - -void ff_float_to_int16_c(int16_t *dst, const float *src, long len){ - int i; - for(i=0; i> 1; + + for (i = 0; i < len2; i++) { + int16_t w = window[i]; + output[i] = (MUL16(input[i], w) + (1 << 14)) >> 15; + output[len-i-1] = (MUL16(input[len-i-1], w) + (1 << 14)) >> 15; + } +} + +static void vector_clip_int32_c(int32_t *dst, const int32_t *src, int32_t min, + int32_t max, unsigned int len) +{ + do { + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + *dst++ = av_clip(*src++, min, max); + len -= 8; + } while (len > 0); +} + #define W0 2048 #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ #define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */ @@ -4397,22 +2703,22 @@ void ff_wmv2_idct_c(short * block){ static void ff_wmv2_idct_put_c(uint8_t *dest, int line_size, DCTELEM *block) { ff_wmv2_idct_c(block); - put_pixels_clamped_c(block, dest, line_size); + ff_put_pixels_clamped_c(block, dest, line_size); } static void ff_wmv2_idct_add_c(uint8_t *dest, int line_size, DCTELEM *block) { ff_wmv2_idct_c(block); - add_pixels_clamped_c(block, dest, line_size); + ff_add_pixels_clamped_c(block, dest, line_size); } static void ff_jref_idct_put(uint8_t *dest, int line_size, DCTELEM *block) { j_rev_dct (block); - put_pixels_clamped_c(block, dest, line_size); + ff_put_pixels_clamped_c(block, dest, line_size); } static void ff_jref_idct_add(uint8_t *dest, int line_size, DCTELEM *block) { j_rev_dct (block); - add_pixels_clamped_c(block, dest, line_size); + ff_add_pixels_clamped_c(block, dest, line_size); } static void ff_jref_idct4_put(uint8_t *dest, int line_size, DCTELEM *block) @@ -4453,7 +2759,7 @@ static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block) static void just_return(void *mem av_unused, int stride av_unused, int h av_unused) { return; } /* init static data */ -void dsputil_static_init(void) +av_cold void dsputil_static_init(void) { int i; @@ -4472,16 +2778,16 @@ void dsputil_static_init(void) int ff_check_alignment(void){ static int did_fail=0; - DECLARE_ALIGNED_16(int, aligned); + LOCAL_ALIGNED_16(int, aligned, [4]); - if((intptr_t)&aligned & 15){ + if((intptr_t)aligned & 15){ if(!did_fail){ #if HAVE_MMX || HAVE_ALTIVEC av_log(NULL, AV_LOG_ERROR, "Compiler did not align stack variables. Libavcodec has been miscompiled\n" "and may be very slow or crash. This is not a bug in libavcodec,\n" "but in the compiler. You may try recompiling using gcc >= 4.2.\n" - "Do not report crashes to FFmpeg developers.\n"); + "Do not report crashes to Libav developers.\n"); #endif did_fail=1; } @@ -4490,35 +2796,35 @@ int ff_check_alignment(void){ return 0; } -void dsputil_init(DSPContext* c, AVCodecContext *avctx) +av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) { int i; ff_check_alignment(); #if CONFIG_ENCODERS - if(avctx->dct_algo==FF_DCT_FASTINT) { - c->fdct = fdct_ifast; - c->fdct248 = fdct_ifast248; - } - else if(avctx->dct_algo==FF_DCT_FAAN) { - c->fdct = ff_faandct; - c->fdct248 = ff_faandct248; - } - else { - c->fdct = ff_jpeg_fdct_islow; //slow/accurate/default - c->fdct248 = ff_fdct248_islow; + if (avctx->bits_per_raw_sample == 10) { + c->fdct = ff_jpeg_fdct_islow_10; + c->fdct248 = ff_fdct248_islow_10; + } else { + if(avctx->dct_algo==FF_DCT_FASTINT) { + c->fdct = fdct_ifast; + c->fdct248 = fdct_ifast248; + } + else if(avctx->dct_algo==FF_DCT_FAAN) { + c->fdct = ff_faandct; + c->fdct248 = ff_faandct248; + } + else { + c->fdct = ff_jpeg_fdct_islow_8; //slow/accurate/default + c->fdct248 = ff_fdct248_islow_8; + } } #endif //CONFIG_ENCODERS if(avctx->lowres==1){ - if(avctx->idct_algo==FF_IDCT_INT || avctx->idct_algo==FF_IDCT_AUTO || !CONFIG_H264_DECODER){ - c->idct_put= ff_jref_idct4_put; - c->idct_add= ff_jref_idct4_add; - }else{ - c->idct_put= ff_h264_lowres_idct_put_c; - c->idct_add= ff_h264_lowres_idct_add_c; - } + c->idct_put= ff_jref_idct4_put; + c->idct_add= ff_jref_idct4_add; c->idct = j_rev_dct4; c->idct_permutation_type= FF_NO_IDCT_PERM; }else if(avctx->lowres==2){ @@ -4532,6 +2838,12 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->idct = j_rev_dct1; c->idct_permutation_type= FF_NO_IDCT_PERM; }else{ + if (avctx->bits_per_raw_sample == 10) { + c->idct_put = ff_simple_idct_put_10; + c->idct_add = ff_simple_idct_add_10; + c->idct = ff_simple_idct_10; + c->idct_permutation_type = FF_NO_IDCT_PERM; + } else { if(avctx->idct_algo==FF_IDCT_INT){ c->idct_put= ff_jref_idct_put; c->idct_add= ff_jref_idct_add; @@ -4557,39 +2869,27 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->idct_put= ff_ea_idct_put_c; c->idct_permutation_type= FF_NO_IDCT_PERM; }else{ //accurate/default - c->idct_put= ff_simple_idct_put; - c->idct_add= ff_simple_idct_add; - c->idct = ff_simple_idct; + c->idct_put = ff_simple_idct_put_8; + c->idct_add = ff_simple_idct_add_8; + c->idct = ff_simple_idct_8; c->idct_permutation_type= FF_NO_IDCT_PERM; } + } } - if (CONFIG_H264_DECODER) { - c->h264_idct_add= ff_h264_idct_add_c; - c->h264_idct8_add= ff_h264_idct8_add_c; - c->h264_idct_dc_add= ff_h264_idct_dc_add_c; - c->h264_idct8_dc_add= ff_h264_idct8_dc_add_c; - c->h264_idct_add16 = ff_h264_idct_add16_c; - c->h264_idct8_add4 = ff_h264_idct8_add4_c; - c->h264_idct_add8 = ff_h264_idct_add8_c; - c->h264_idct_add16intra= ff_h264_idct_add16intra_c; - } - - c->get_pixels = get_pixels_c; c->diff_pixels = diff_pixels_c; - c->put_pixels_clamped = put_pixels_clamped_c; - c->put_signed_pixels_clamped = put_signed_pixels_clamped_c; - c->add_pixels_clamped = add_pixels_clamped_c; - c->add_pixels8 = add_pixels8_c; - c->add_pixels4 = add_pixels4_c; + c->put_pixels_clamped = ff_put_pixels_clamped_c; + c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_c; + c->add_pixels_clamped = ff_add_pixels_clamped_c; c->sum_abs_dctelem = sum_abs_dctelem_c; c->gmc1 = gmc1_c; c->gmc = ff_gmc_c; - c->clear_block = clear_block_c; - c->clear_blocks = clear_blocks_c; c->pix_sum = pix_sum_c; c->pix_norm1 = pix_norm1_c; + c->fill_block_tab[0] = fill_block16_c; + c->fill_block_tab[1] = fill_block8_c; + /* TODO [0] 16 [1] 8 */ c->pix_abs[0][0] = pix_abs16_c; c->pix_abs[0][1] = pix_abs16_x2_c; @@ -4600,30 +2900,6 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->pix_abs[1][2] = pix_abs8_y2_c; c->pix_abs[1][3] = pix_abs8_xy2_c; -#define dspfunc(PFX, IDX, NUM) \ - c->PFX ## _pixels_tab[IDX][0] = PFX ## _pixels ## NUM ## _c; \ - c->PFX ## _pixels_tab[IDX][1] = PFX ## _pixels ## NUM ## _x2_c; \ - c->PFX ## _pixels_tab[IDX][2] = PFX ## _pixels ## NUM ## _y2_c; \ - c->PFX ## _pixels_tab[IDX][3] = PFX ## _pixels ## NUM ## _xy2_c - - dspfunc(put, 0, 16); - dspfunc(put_no_rnd, 0, 16); - dspfunc(put, 1, 8); - dspfunc(put_no_rnd, 1, 8); - dspfunc(put, 2, 4); - dspfunc(put, 3, 2); - - dspfunc(avg, 0, 16); - dspfunc(avg_no_rnd, 0, 16); - dspfunc(avg, 1, 8); - dspfunc(avg_no_rnd, 1, 8); - dspfunc(avg, 2, 4); - dspfunc(avg, 3, 2); -#undef dspfunc - - c->put_no_rnd_pixels_l2[0]= put_no_rnd_pixels16_l2_c; - c->put_no_rnd_pixels_l2[1]= put_no_rnd_pixels8_l2_c; - c->put_tpel_pixels_tab[ 0] = put_tpel_pixels_mc00_c; c->put_tpel_pixels_tab[ 1] = put_tpel_pixels_mc10_c; c->put_tpel_pixels_tab[ 2] = put_tpel_pixels_mc20_c; @@ -4674,72 +2950,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) dspfunc(avg_qpel, 1, 8); /* dspfunc(avg_no_rnd_qpel, 1, 8); */ - dspfunc(put_h264_qpel, 0, 16); - dspfunc(put_h264_qpel, 1, 8); - dspfunc(put_h264_qpel, 2, 4); - dspfunc(put_h264_qpel, 3, 2); - dspfunc(avg_h264_qpel, 0, 16); - dspfunc(avg_h264_qpel, 1, 8); - dspfunc(avg_h264_qpel, 2, 4); - #undef dspfunc - c->put_h264_chroma_pixels_tab[0]= put_h264_chroma_mc8_c; - c->put_h264_chroma_pixels_tab[1]= put_h264_chroma_mc4_c; - c->put_h264_chroma_pixels_tab[2]= put_h264_chroma_mc2_c; - c->avg_h264_chroma_pixels_tab[0]= avg_h264_chroma_mc8_c; - c->avg_h264_chroma_pixels_tab[1]= avg_h264_chroma_mc4_c; - c->avg_h264_chroma_pixels_tab[2]= avg_h264_chroma_mc2_c; - c->put_no_rnd_vc1_chroma_pixels_tab[0]= put_no_rnd_vc1_chroma_mc8_c; - c->avg_no_rnd_vc1_chroma_pixels_tab[0]= avg_no_rnd_vc1_chroma_mc8_c; - - c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c; - c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c; - c->weight_h264_pixels_tab[2]= weight_h264_pixels8x16_c; - c->weight_h264_pixels_tab[3]= weight_h264_pixels8x8_c; - c->weight_h264_pixels_tab[4]= weight_h264_pixels8x4_c; - c->weight_h264_pixels_tab[5]= weight_h264_pixels4x8_c; - c->weight_h264_pixels_tab[6]= weight_h264_pixels4x4_c; - c->weight_h264_pixels_tab[7]= weight_h264_pixels4x2_c; - c->weight_h264_pixels_tab[8]= weight_h264_pixels2x4_c; - c->weight_h264_pixels_tab[9]= weight_h264_pixels2x2_c; - c->biweight_h264_pixels_tab[0]= biweight_h264_pixels16x16_c; - c->biweight_h264_pixels_tab[1]= biweight_h264_pixels16x8_c; - c->biweight_h264_pixels_tab[2]= biweight_h264_pixels8x16_c; - c->biweight_h264_pixels_tab[3]= biweight_h264_pixels8x8_c; - c->biweight_h264_pixels_tab[4]= biweight_h264_pixels8x4_c; - c->biweight_h264_pixels_tab[5]= biweight_h264_pixels4x8_c; - c->biweight_h264_pixels_tab[6]= biweight_h264_pixels4x4_c; - c->biweight_h264_pixels_tab[7]= biweight_h264_pixels4x2_c; - c->biweight_h264_pixels_tab[8]= biweight_h264_pixels2x4_c; - c->biweight_h264_pixels_tab[9]= biweight_h264_pixels2x2_c; - - c->draw_edges = draw_edges_c; - -#if CONFIG_CAVS_DECODER - ff_cavsdsp_init(c,avctx); -#endif #if CONFIG_MLP_DECODER || CONFIG_TRUEHD_DECODER ff_mlp_init(c, avctx); #endif -#if CONFIG_VC1_DECODER - ff_vc1dsp_init(c,avctx); -#endif #if CONFIG_WMV2_DECODER || CONFIG_VC1_DECODER ff_intrax8dsp_init(c,avctx); #endif -#if CONFIG_RV30_DECODER - ff_rv30dsp_init(c,avctx); -#endif -#if CONFIG_RV40_DECODER - ff_rv40dsp_init(c,avctx); - c->put_rv40_qpel_pixels_tab[0][15] = put_rv40_qpel16_mc33_c; - c->avg_rv40_qpel_pixels_tab[0][15] = avg_rv40_qpel16_mc33_c; - c->put_rv40_qpel_pixels_tab[1][15] = put_rv40_qpel8_mc33_c; - c->avg_rv40_qpel_pixels_tab[1][15] = avg_rv40_qpel8_mc33_c; -#endif - c->put_mspel_pixels_tab[0]= put_mspel8_mc00_c; + c->put_mspel_pixels_tab[0]= ff_put_pixels8x8_c; c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c; c->put_mspel_pixels_tab[2]= put_mspel8_mc20_c; c->put_mspel_pixels_tab[3]= put_mspel8_mc30_c; @@ -4776,11 +2996,8 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->vsse[5]= vsse_intra8_c; c->nsse[0]= nsse16_c; c->nsse[1]= nsse8_c; -#if CONFIG_SNOW_ENCODER - c->w53[0]= w53_16_c; - c->w53[1]= w53_8_c; - c->w97[0]= w97_16_c; - c->w97[1]= w97_8_c; +#if CONFIG_DWT + ff_dsputil_init_dwt(c); #endif c->ssd_int8_vs_int16 = ssd_int8_vs_int16_c; @@ -4793,21 +3010,12 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->add_hfyu_left_prediction = add_hfyu_left_prediction_c; c->add_hfyu_left_prediction_bgr32 = add_hfyu_left_prediction_bgr32_c; c->bswap_buf= bswap_buf; + c->bswap16_buf = bswap16_buf; #if CONFIG_PNG_DECODER c->add_png_paeth_prediction= ff_add_png_paeth_prediction; #endif - c->h264_v_loop_filter_luma= h264_v_loop_filter_luma_c; - c->h264_h_loop_filter_luma= h264_h_loop_filter_luma_c; - c->h264_v_loop_filter_luma_intra= h264_v_loop_filter_luma_intra_c; - c->h264_h_loop_filter_luma_intra= h264_h_loop_filter_luma_intra_c; - c->h264_v_loop_filter_chroma= h264_v_loop_filter_chroma_c; - c->h264_h_loop_filter_chroma= h264_h_loop_filter_chroma_c; - c->h264_v_loop_filter_chroma_intra= h264_v_loop_filter_chroma_intra_c; - c->h264_h_loop_filter_chroma_intra= h264_h_loop_filter_chroma_intra_c; - c->h264_loop_filter_strength= NULL; - - if (CONFIG_ANY_H263) { + if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) { c->h263_h_loop_filter= h263_h_loop_filter_c; c->h263_v_loop_filter= h263_v_loop_filter_c; } @@ -4815,9 +3023,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) if (CONFIG_VP3_DECODER) { c->vp3_h_loop_filter= ff_vp3_h_loop_filter_c; c->vp3_v_loop_filter= ff_vp3_v_loop_filter_c; - } - if (CONFIG_VP6_DECODER) { - c->vp6_filter_diag4= ff_vp6_filter_diag4_c; + c->vp3_idct_dc_add= ff_vp3_idct_dc_add_c; } c->h261_loop_filter= h261_loop_filter_c; @@ -4825,42 +3031,28 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->try_8x8basis= try_8x8basis_c; c->add_8x8basis= add_8x8basis_c; -#if CONFIG_SNOW_DECODER - c->vertical_compose97i = ff_snow_vertical_compose97i; - c->horizontal_compose97i = ff_snow_horizontal_compose97i; - c->inner_add_yblock = ff_snow_inner_add_yblock; -#endif - #if CONFIG_VORBIS_DECODER c->vorbis_inverse_coupling = vorbis_inverse_coupling; #endif #if CONFIG_AC3_DECODER c->ac3_downmix = ff_ac3_downmix_c; -#endif -#if CONFIG_LPC - c->lpc_compute_autocorr = ff_lpc_compute_autocorr; #endif c->vector_fmul = vector_fmul_c; c->vector_fmul_reverse = vector_fmul_reverse_c; c->vector_fmul_add = vector_fmul_add_c; - c->vector_fmul_window = ff_vector_fmul_window_c; - c->int32_to_float_fmul_scalar = int32_to_float_fmul_scalar_c; + c->vector_fmul_window = vector_fmul_window_c; c->vector_clipf = vector_clipf_c; - c->float_to_int16 = ff_float_to_int16_c; - c->float_to_int16_interleave = ff_float_to_int16_interleave_c; c->scalarproduct_int16 = scalarproduct_int16_c; c->scalarproduct_and_madd_int16 = scalarproduct_and_madd_int16_c; + c->apply_window_int16 = apply_window_int16_c; + c->vector_clip_int32 = vector_clip_int32_c; c->scalarproduct_float = scalarproduct_float_c; c->butterflies_float = butterflies_float_c; + c->butterflies_float_interleave = butterflies_float_interleave_c; c->vector_fmul_scalar = vector_fmul_scalar_c; + c->vector_fmac_scalar = vector_fmac_scalar_c; - c->vector_fmul_sv_scalar[0] = vector_fmul_sv_scalar_2_c; - c->vector_fmul_sv_scalar[1] = vector_fmul_sv_scalar_4_c; - - c->sv_fmul_scalar[0] = sv_fmul_scalar_2_c; - c->sv_fmul_scalar[1] = sv_fmul_scalar_4_c; - - c->shrink[0]= ff_img_copy_plane; + c->shrink[0]= av_image_copy_plane; c->shrink[1]= ff_shrink22; c->shrink[2]= ff_shrink44; c->shrink[3]= ff_shrink88; @@ -4870,6 +3062,98 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) memset(c->put_2tap_qpel_pixels_tab, 0, sizeof(c->put_2tap_qpel_pixels_tab)); memset(c->avg_2tap_qpel_pixels_tab, 0, sizeof(c->avg_2tap_qpel_pixels_tab)); +#undef FUNC +#undef FUNCC +#define FUNC(f, depth) f ## _ ## depth +#define FUNCC(f, depth) f ## _ ## depth ## _c + +#define dspfunc1(PFX, IDX, NUM, depth)\ + c->PFX ## _pixels_tab[IDX][0] = FUNCC(PFX ## _pixels ## NUM , depth);\ + c->PFX ## _pixels_tab[IDX][1] = FUNCC(PFX ## _pixels ## NUM ## _x2 , depth);\ + c->PFX ## _pixels_tab[IDX][2] = FUNCC(PFX ## _pixels ## NUM ## _y2 , depth);\ + c->PFX ## _pixels_tab[IDX][3] = FUNCC(PFX ## _pixels ## NUM ## _xy2, depth) + +#define dspfunc2(PFX, IDX, NUM, depth)\ + c->PFX ## _pixels_tab[IDX][ 0] = FUNCC(PFX ## NUM ## _mc00, depth);\ + c->PFX ## _pixels_tab[IDX][ 1] = FUNCC(PFX ## NUM ## _mc10, depth);\ + c->PFX ## _pixels_tab[IDX][ 2] = FUNCC(PFX ## NUM ## _mc20, depth);\ + c->PFX ## _pixels_tab[IDX][ 3] = FUNCC(PFX ## NUM ## _mc30, depth);\ + c->PFX ## _pixels_tab[IDX][ 4] = FUNCC(PFX ## NUM ## _mc01, depth);\ + c->PFX ## _pixels_tab[IDX][ 5] = FUNCC(PFX ## NUM ## _mc11, depth);\ + c->PFX ## _pixels_tab[IDX][ 6] = FUNCC(PFX ## NUM ## _mc21, depth);\ + c->PFX ## _pixels_tab[IDX][ 7] = FUNCC(PFX ## NUM ## _mc31, depth);\ + c->PFX ## _pixels_tab[IDX][ 8] = FUNCC(PFX ## NUM ## _mc02, depth);\ + c->PFX ## _pixels_tab[IDX][ 9] = FUNCC(PFX ## NUM ## _mc12, depth);\ + c->PFX ## _pixels_tab[IDX][10] = FUNCC(PFX ## NUM ## _mc22, depth);\ + c->PFX ## _pixels_tab[IDX][11] = FUNCC(PFX ## NUM ## _mc32, depth);\ + c->PFX ## _pixels_tab[IDX][12] = FUNCC(PFX ## NUM ## _mc03, depth);\ + c->PFX ## _pixels_tab[IDX][13] = FUNCC(PFX ## NUM ## _mc13, depth);\ + c->PFX ## _pixels_tab[IDX][14] = FUNCC(PFX ## NUM ## _mc23, depth);\ + c->PFX ## _pixels_tab[IDX][15] = FUNCC(PFX ## NUM ## _mc33, depth) + + +#define BIT_DEPTH_FUNCS(depth, dct)\ + c->get_pixels = FUNCC(get_pixels ## dct , depth);\ + c->draw_edges = FUNCC(draw_edges , depth);\ + c->emulated_edge_mc = FUNC (ff_emulated_edge_mc , depth);\ + c->clear_block = FUNCC(clear_block ## dct , depth);\ + c->clear_blocks = FUNCC(clear_blocks ## dct , depth);\ + c->add_pixels8 = FUNCC(add_pixels8 ## dct , depth);\ + c->add_pixels4 = FUNCC(add_pixels4 ## dct , depth);\ + c->put_no_rnd_pixels_l2[0] = FUNCC(put_no_rnd_pixels16_l2, depth);\ + c->put_no_rnd_pixels_l2[1] = FUNCC(put_no_rnd_pixels8_l2 , depth);\ +\ + c->put_h264_chroma_pixels_tab[0] = FUNCC(put_h264_chroma_mc8 , depth);\ + c->put_h264_chroma_pixels_tab[1] = FUNCC(put_h264_chroma_mc4 , depth);\ + c->put_h264_chroma_pixels_tab[2] = FUNCC(put_h264_chroma_mc2 , depth);\ + c->avg_h264_chroma_pixels_tab[0] = FUNCC(avg_h264_chroma_mc8 , depth);\ + c->avg_h264_chroma_pixels_tab[1] = FUNCC(avg_h264_chroma_mc4 , depth);\ + c->avg_h264_chroma_pixels_tab[2] = FUNCC(avg_h264_chroma_mc2 , depth);\ +\ + dspfunc1(put , 0, 16, depth);\ + dspfunc1(put , 1, 8, depth);\ + dspfunc1(put , 2, 4, depth);\ + dspfunc1(put , 3, 2, depth);\ + dspfunc1(put_no_rnd, 0, 16, depth);\ + dspfunc1(put_no_rnd, 1, 8, depth);\ + dspfunc1(avg , 0, 16, depth);\ + dspfunc1(avg , 1, 8, depth);\ + dspfunc1(avg , 2, 4, depth);\ + dspfunc1(avg , 3, 2, depth);\ + dspfunc1(avg_no_rnd, 0, 16, depth);\ + dspfunc1(avg_no_rnd, 1, 8, depth);\ +\ + dspfunc2(put_h264_qpel, 0, 16, depth);\ + dspfunc2(put_h264_qpel, 1, 8, depth);\ + dspfunc2(put_h264_qpel, 2, 4, depth);\ + dspfunc2(put_h264_qpel, 3, 2, depth);\ + dspfunc2(avg_h264_qpel, 0, 16, depth);\ + dspfunc2(avg_h264_qpel, 1, 8, depth);\ + dspfunc2(avg_h264_qpel, 2, 4, depth); + + switch (avctx->bits_per_raw_sample) { + case 9: + if (c->dct_bits == 32) { + BIT_DEPTH_FUNCS(9, _32); + } else { + BIT_DEPTH_FUNCS(9, _16); + } + break; + case 10: + if (c->dct_bits == 32) { + BIT_DEPTH_FUNCS(10, _32); + } else { + BIT_DEPTH_FUNCS(10, _16); + } + break; + default: + av_log(avctx, AV_LOG_DEBUG, "Unsupported bit depth: %d\n", avctx->bits_per_raw_sample); + case 8: + BIT_DEPTH_FUNCS(8, _16); + break; + } + + if (HAVE_MMX) dsputil_init_mmx (c, avctx); if (ARCH_ARM) dsputil_init_arm (c, avctx); if (CONFIG_MLIB) dsputil_init_mlib (c, avctx); @@ -4887,33 +3171,6 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i]; } - switch(c->idct_permutation_type){ - case FF_NO_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= i; - break; - case FF_LIBMPEG2_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2); - break; - case FF_SIMPLE_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= simple_mmx_permutation[i]; - break; - case FF_TRANSPOSE_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= ((i&7)<<3) | (i>>3); - break; - case FF_PARTTRANS_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= (i&0x24) | ((i&3)<<3) | ((i>>3)&3); - break; - case FF_SSE2_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= (i&0x38) | idct_sse2_row_perm[i&7]; - break; - default: - av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n"); - } + ff_init_scantable_permutation(c->idct_permutation, + c->idct_permutation_type); } -