X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.c;h=f42c58516f79ce3c0c78a3e0a07229c40769c064;hb=37ddd3833219fa7b913fff3f5cccc6878b047e6b;hp=84f91fc135b9e7712a44c473a42b1c0f08813ea6;hpb=5ce5dbc5f3d0bce1f8d76fea1907c91469ebdd01;p=ffmpeg diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index 84f91fc135b..f42c58516f7 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -5,20 +5,20 @@ * * 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 */ @@ -27,7 +27,7 @@ * DSP utils */ -#include "libavcore/imgutils.h" +#include "libavutil/imgutils.h" #include "avcodec.h" #include "dsputil.h" #include "simple_idct.h" @@ -38,11 +38,21 @@ #include "config.h" #include "ac3dec.h" #include "vorbis.h" -#include "png.h" uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, }; uint32_t ff_squareTbl[512] = {0, }; +#define BIT_DEPTH 9 +#include "dsputil_template.c" +#undef BIT_DEPTH + +#define BIT_DEPTH 10 +#include "dsputil_template.c" +#undef BIT_DEPTH + +#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) #define pb_80 (~0UL/255 * 0x80) @@ -72,7 +82,7 @@ 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, ff_inv_zigzag_direct16)[64]; const uint8_t ff_alternate_horizontal_scan[64] = { 0, 1, 2, 3, 8, 9, 16, 17, @@ -120,9 +130,6 @@ void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_s int j; j = src_scantable[i]; st->permutated[i] = permutation[j]; -#if ARCH_PPC - st->inverse[j] = i; -#endif } end=-1; @@ -134,6 +141,41 @@ 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; @@ -174,7 +216,7 @@ 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]; @@ -222,6 +264,12 @@ static void bswap_buf(uint32_t *dst, const uint32_t *src, int w){ } } +static void bswap16_buf(uint16_t *dst, const uint16_t *src, int len) +{ + while (len--) + *dst++ = av_bswap16(*src++); +} + static int sse4_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h) { int s, i; @@ -290,125 +338,6 @@ static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) return s; } -/* 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, const uint8_t *src2, const uint8_t *src3, const 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, const uint8_t *src2, const uint8_t *src3, const 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, const uint8_t *src2, const uint8_t *src3, const 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)\ -av_unused 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 - -PIXOP2(avg, op_avg) -PIXOP2(put, op_put) -#undef op_avg -#undef op_put - -#define put_no_rnd_pixels8_c put_pixels8_c -#define put_no_rnd_pixels16_c put_pixels16_c - #define avg2(a,b) ((a+b+1)>>1) #define avg4(a,b,c,d) ((a+b+c+d+2)>>2) -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); -} - -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); -} - static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y16, int rounder) { const int A=(16-x16)*(16-y16); @@ -1271,10 +535,10 @@ 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; } } @@ -1368,10 +632,10 @@ 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; } } @@ -1423,221 +687,43 @@ static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int int i,j; for (i=0; i < height; i++) { for (j=0; j < width; j++) { - dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1; - } - src += stride; - dst += stride; - } -} - -static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ - int i,j; - for (i=0; i < height; i++) { - for (j=0; j < width; j++) { - dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1; - } - src += stride; - dst += stride; - } -} - -static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ - int i,j; - for (i=0; i < height; i++) { - for (j=0; j < width; j++) { - dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1; - } - src += stride; - dst += stride; - } -} - -static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ - int i,j; - for (i=0; i < height; i++) { - for (j=0; j < width; j++) { - dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1; - } - src += stride; - 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; + dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1; + } + src += stride; + dst += 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; +static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ + int i,j; + for (i=0; i < height; i++) { + for (j=0; j < width; j++) { + dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1; + } + src += stride; + dst += stride; + } +} - assert(x<8 && y<8 && x>=0 && y>=0); +static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ + int i,j; + for (i=0; i < height; i++) { + for (j=0; j < width; j++) { + dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1; + } + src += stride; + dst += stride; + } +} - 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; +static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){ + int i,j; + for (i=0; i < height; i++) { + for (j=0; j < width; j++) { + dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1; + } + src += stride; + dst += stride; } } @@ -1762,7 +848,7 @@ static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dst 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){\ @@ -1772,7 +858,7 @@ 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){\ @@ -1780,7 +866,7 @@ 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){\ @@ -1794,7 +880,7 @@ 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];\ @@ -1805,7 +891,7 @@ 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];\ @@ -1813,9 +899,9 @@ 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];\ @@ -1826,7 +912,7 @@ 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];\ @@ -1834,9 +920,9 @@ 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];\ @@ -1847,7 +933,7 @@ 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];\ @@ -1855,9 +941,9 @@ 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];\ @@ -1868,7 +954,7 @@ 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];\ @@ -1876,23 +962,23 @@ 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];\ @@ -1903,14 +989,14 @@ 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){\ @@ -1922,14 +1008,14 @@ 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){\ @@ -1941,7 +1027,7 @@ static void OPNAME ## qpel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){\ 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){\ @@ -1951,7 +1037,7 @@ 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){\ @@ -1959,7 +1045,7 @@ 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){\ @@ -1973,7 +1059,7 @@ 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];\ @@ -1984,7 +1070,7 @@ 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];\ @@ -1992,9 +1078,9 @@ 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];\ @@ -2005,567 +1091,138 @@ 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];\ - 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, 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];\ - 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_l4(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];\ - 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, 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);\ -}\ -void ff_ ## OPNAME ## qpel16_mc33_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_l4(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 - -#define put_qpel8_mc00_c ff_put_pixels8x8_c -#define avg_qpel8_mc00_c ff_avg_pixels8x8_c -#define put_qpel16_mc00_c ff_put_pixels16x16_c -#define avg_qpel16_mc00_c ff_avg_pixels16x16_c -#define put_no_rnd_qpel8_mc00_c ff_put_pixels8x8_c -#define put_no_rnd_qpel16_mc00_c ff_put_pixels16x16_c - -#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 +#undef op_put_no_rnd -#define put_h264_qpel8_mc00_c ff_put_pixels8x8_c -#define avg_h264_qpel8_mc00_c ff_avg_pixels8x8_c -#define put_h264_qpel16_mc00_c ff_put_pixels16x16_c -#define avg_h264_qpel16_mc00_c ff_avg_pixels16x16_c +#define put_qpel8_mc00_c ff_put_pixels8x8_c +#define avg_qpel8_mc00_c ff_avg_pixels8x8_c +#define put_qpel16_mc00_c ff_put_pixels16x16_c +#define avg_qpel16_mc00_c ff_avg_pixels16x16_c +#define put_no_rnd_qpel8_mc00_c ff_put_pixels8x8_c +#define put_no_rnd_qpel16_mc00_c ff_put_pixels16x16_8_c static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){ uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; @@ -2585,31 +1242,18 @@ static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int } } -void ff_put_pixels8x8_c(uint8_t *dst, uint8_t *src, int stride) { - put_pixels8_c(dst, src, stride, 8); -} -void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, int stride) { - avg_pixels8_c(dst, src, stride, 8); -} -void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, int stride) { - put_pixels16_c(dst, src, stride, 16); -} -void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, int stride) { - avg_pixels16_c(dst, src, stride, 16); -} - #if CONFIG_RV40_DECODER -static void put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ - put_pixels16_xy2_c(dst, src, stride, 16); +void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ + put_pixels16_xy2_8_c(dst, src, stride, 16); } -static void avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ - avg_pixels16_xy2_c(dst, src, stride, 16); +void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ + avg_pixels16_xy2_8_c(dst, src, stride, 16); } -static void put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ - put_pixels8_xy2_c(dst, src, stride, 8); +void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ + put_pixels8_xy2_8_c(dst, src, stride, 8); } -static void avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ - avg_pixels8_xy2_c(dst, src, stride, 8); +void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ + avg_pixels8_xy2_8_c(dst, src, stride, 8); } #endif /* CONFIG_RV40_DECODER */ @@ -2645,7 +1289,7 @@ static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int static void put_mspel8_mc10_c(uint8_t *dst, uint8_t *src, int stride){ uint8_t half[64]; wmv2_mspel8_h_lowpass(half, src, 8, stride, 8); - put_pixels8_l2(dst, src, half, stride, stride, 8, 8); + put_pixels8_l2_8(dst, src, half, stride, stride, 8, 8); } static void put_mspel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){ @@ -2655,7 +1299,7 @@ static void put_mspel8_mc20_c(uint8_t *dst, uint8_t *src, int stride){ static void put_mspel8_mc30_c(uint8_t *dst, uint8_t *src, int stride){ uint8_t half[64]; wmv2_mspel8_h_lowpass(half, src, 8, stride, 8); - put_pixels8_l2(dst, src+1, half, stride, stride, 8, 8); + put_pixels8_l2_8(dst, src+1, half, stride, stride, 8, 8); } static void put_mspel8_mc02_c(uint8_t *dst, uint8_t *src, int stride){ @@ -2669,7 +1313,7 @@ static void put_mspel8_mc12_c(uint8_t *dst, uint8_t *src, int stride){ wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11); wmv2_mspel8_v_lowpass(halfV, src, 8, stride, 8); wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8); - put_pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8); + put_pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); } static void put_mspel8_mc32_c(uint8_t *dst, uint8_t *src, int stride){ uint8_t halfH[88]; @@ -2678,7 +1322,7 @@ static void put_mspel8_mc32_c(uint8_t *dst, uint8_t *src, int stride){ wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11); wmv2_mspel8_v_lowpass(halfV, src+1, 8, stride, 8); wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8); - put_pixels8_l2(dst, halfV, halfHV, stride, 8, 8, 8); + put_pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8); } static void put_mspel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){ uint8_t halfH[88]; @@ -3063,7 +1707,7 @@ 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 @@ -3155,19 +1799,6 @@ 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)){ @@ -3179,17 +1810,6 @@ static void add_bytes_c(uint8_t *dst, uint8_t *src, int w){ dst[i+0] += src[i+0]; } -static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){ - long i; - for(i=0; i<=w-sizeof(long); i+=sizeof(long)){ - long a = *(long*)(src1+i); - long b = *(long*)(src2+i); - *(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80); - } - for(; imaxi){ - maxi=sum; - printf("MAX:%d\n", maxi); -} -#endif return sum; } @@ -3520,7 +2133,7 @@ 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]); @@ -3769,7 +2382,9 @@ static void vector_fmul_add_c(float *dst, const float *src0, const float *src1, dst[i] = src0[i] * src1[i] + src2[i]; } -void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1, const float *win, float add_bias, int len){ +static void vector_fmul_window_c(float *dst, const float *src0, + const float *src1, const float *win, int len) +{ int i,j; dst += len; win += len; @@ -3779,8 +2394,8 @@ void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1, c float s1 = src1[j]; float wi = win[i]; float wj = win[j]; - dst[i] = s0*wj - s1*wi + add_bias; - dst[j] = s0*wi + s1*wj + add_bias; + dst[i] = s0*wj - s1*wi; + dst[j] = s0*wi + s1*wj; } } @@ -3792,48 +2407,12 @@ static void vector_fmul_scalar_c(float *dst, const float *src, float mul, dst[i] = src[i] * mul; } -static void vector_fmul_sv_scalar_2_c(float *dst, const float *src, - const float **sv, float mul, int len) -{ - int i; - for (i = 0; i < len; i += 2, sv++) { - dst[i ] = src[i ] * sv[0][0] * mul; - dst[i+1] = src[i+1] * sv[0][1] * mul; - } -} - -static void vector_fmul_sv_scalar_4_c(float *dst, const float *src, - const float **sv, float mul, int len) -{ - int i; - for (i = 0; i < len; i += 4, sv++) { - dst[i ] = src[i ] * sv[0][0] * mul; - dst[i+1] = src[i+1] * sv[0][1] * mul; - dst[i+2] = src[i+2] * sv[0][2] * mul; - dst[i+3] = src[i+3] * sv[0][3] * mul; - } -} - -static void sv_fmul_scalar_2_c(float *dst, const float **sv, float mul, - int len) -{ - int i; - for (i = 0; i < len; i += 2, sv++) { - dst[i ] = sv[0][0] * mul; - dst[i+1] = sv[0][1] * mul; - } -} - -static void sv_fmul_scalar_4_c(float *dst, const float **sv, float mul, - int len) +static void vector_fmac_scalar_c(float *dst, const float *src, float mul, + int len) { int i; - for (i = 0; i < len; i += 4, sv++) { - dst[i ] = sv[0][0] * mul; - dst[i+1] = sv[0][1] * mul; - dst[i+2] = sv[0][2] * mul; - dst[i+3] = sv[0][3] * mul; - } + for (i = 0; i < len; i++) + dst[i] += src[i] * mul; } static void butterflies_float_c(float *restrict v1, float *restrict v2, @@ -3847,6 +2426,18 @@ static void butterflies_float_c(float *restrict v1, float *restrict v2, } } +static void butterflies_float_interleave_c(float *dst, const float *src0, + const float *src1, int len) +{ + int i; + for (i = 0; i < len; i++) { + float f1 = src0[i]; + float f2 = src1[i]; + dst[2*i ] = f1 + f2; + dst[2*i + 1] = f1 - f2; + } +} + static float scalarproduct_float_c(const float *v1, const float *v2, int len) { float p = 0.0; @@ -3858,18 +2449,12 @@ static float scalarproduct_float_c(const float *v1, const float *v2, int len) return p; } -static void int32_to_float_fmul_scalar_c(float *dst, const int *src, float mul, 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; } @@ -3877,7 +2462,7 @@ 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> shift; + res += *v1++ * *v2++; return res; } @@ -3953,6 +2514,35 @@ static int32_t scalarproduct_and_madd_int16_c(int16_t *v1, const int16_t *v2, co return res; } +static void apply_window_int16_c(int16_t *output, const int16_t *input, + const int16_t *window, unsigned int len) +{ + int i; + int len2 = len >> 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) */ @@ -4030,63 +2620,28 @@ 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_j_rev_dct (block); + 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); -} - -static void ff_jref_idct4_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct4 (block); - put_pixels_clamped4_c(block, dest, line_size); -} -static void ff_jref_idct4_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct4 (block); - add_pixels_clamped4_c(block, dest, line_size); -} - -static void ff_jref_idct2_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct2 (block); - put_pixels_clamped2_c(block, dest, line_size); -} -static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - j_rev_dct2 (block); - add_pixels_clamped2_c(block, dest, line_size); -} - -static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block) -{ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - dest[0] = cm[(block[0] + 4)>>3]; -} -static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - - dest[0] = cm[dest[0] + ((block[0] + 4)>>3)]; + ff_j_rev_dct (block); + ff_add_pixels_clamped_c(block, dest, line_size); } static void just_return(void *mem av_unused, int stride av_unused, int h av_unused) { return; } /* init static data */ -av_cold void dsputil_static_init(void) +av_cold void ff_dsputil_static_init(void) { int i; @@ -4100,21 +2655,21 @@ av_cold void dsputil_static_init(void) ff_squareTbl[i] = (i - 256) * (i - 256); } - for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1; + for(i=0; i<64; i++) ff_inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1; } 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; } @@ -4123,52 +2678,42 @@ int ff_check_alignment(void){ return 0; } -av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) +av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx) { - int i; + int i, j; 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 = ff_fdct_ifast; + c->fdct248 = ff_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 = j_rev_dct4; - c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(avctx->lowres==2){ - c->idct_put= ff_jref_idct2_put; - c->idct_add= ff_jref_idct2_add; - c->idct = j_rev_dct2; - c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(avctx->lowres==3){ - c->idct_put= ff_jref_idct1_put; - c->idct_add= ff_jref_idct1_add; - 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; - c->idct = j_rev_dct; + c->idct = ff_j_rev_dct; c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM; }else if((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER || CONFIG_VP6_DECODER ) && avctx->idct_algo==FF_IDCT_VP3){ @@ -4189,38 +2734,26 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) }else if(CONFIG_EATGQ_DECODER && avctx->idct_algo==FF_IDCT_EA) { c->idct_put= ff_ea_idct_put_c; c->idct_permutation_type= FF_NO_IDCT_PERM; - }else if(CONFIG_BINK_DECODER && avctx->idct_algo==FF_IDCT_BINK) { - c->idct = ff_bink_idct_c; - c->idct_add = ff_bink_idct_add_c; - c->idct_put = ff_bink_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; } } - 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->put_pixels_nonclamped = put_pixels_nonclamped_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; - c->scale_block = scale_block_c; /* TODO [0] 16 [1] 8 */ c->pix_abs[0][0] = pix_abs16_c; @@ -4232,30 +2765,6 @@ av_cold 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; @@ -4306,45 +2815,14 @@ av_cold 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->draw_edges = draw_edges_c; #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]= ff_put_pixels8x8_c; c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c; @@ -4390,16 +2868,13 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->ssd_int8_vs_int16 = ssd_int8_vs_int16_c; c->add_bytes= add_bytes_c; - c->add_bytes_l2= add_bytes_l2_c; c->diff_bytes= diff_bytes_c; c->add_hfyu_median_prediction= add_hfyu_median_prediction_c; c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_c; 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; -#if CONFIG_PNG_DECODER - c->add_png_paeth_prediction= ff_add_png_paeth_prediction; -#endif + c->bswap16_buf = bswap16_buf; if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) { c->h263_h_loop_filter= h263_h_loop_filter_c; @@ -4418,7 +2893,7 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) c->add_8x8basis= add_8x8basis_c; #if CONFIG_VORBIS_DECODER - c->vorbis_inverse_coupling = vorbis_inverse_coupling; + c->vorbis_inverse_coupling = ff_vorbis_inverse_coupling; #endif #if CONFIG_AC3_DECODER c->ac3_downmix = ff_ac3_downmix_c; @@ -4426,22 +2901,17 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) 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_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->vector_fmac_scalar = vector_fmac_scalar_c; c->shrink[0]= av_image_copy_plane; c->shrink[1]= ff_shrink22; @@ -4453,60 +2923,116 @@ av_cold 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)); - if (HAVE_MMX) dsputil_init_mmx (c, avctx); - if (ARCH_ARM) dsputil_init_arm (c, avctx); - if (CONFIG_MLIB) dsputil_init_mlib (c, avctx); - if (HAVE_VIS) dsputil_init_vis (c, avctx); - if (ARCH_ALPHA) dsputil_init_alpha (c, avctx); - if (ARCH_PPC) dsputil_init_ppc (c, avctx); - if (HAVE_MMI) dsputil_init_mmi (c, avctx); - if (ARCH_SH4) dsputil_init_sh4 (c, avctx); - if (ARCH_BFIN) dsputil_init_bfin (c, avctx); - - for(i=0; i<64; i++){ - if(!c->put_2tap_qpel_pixels_tab[0][i]) - c->put_2tap_qpel_pixels_tab[0][i]= c->put_h264_qpel_pixels_tab[0][i]; - if(!c->avg_2tap_qpel_pixels_tab[0][i]) - c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i]; - } - - c->put_rv30_tpel_pixels_tab[0][0] = c->put_h264_qpel_pixels_tab[0][0]; - c->put_rv30_tpel_pixels_tab[1][0] = c->put_h264_qpel_pixels_tab[1][0]; - c->avg_rv30_tpel_pixels_tab[0][0] = c->avg_h264_qpel_pixels_tab[0][0]; - c->avg_rv30_tpel_pixels_tab[1][0] = c->avg_h264_qpel_pixels_tab[1][0]; - - c->put_rv40_qpel_pixels_tab[0][0] = c->put_h264_qpel_pixels_tab[0][0]; - c->put_rv40_qpel_pixels_tab[1][0] = c->put_h264_qpel_pixels_tab[1][0]; - c->avg_rv40_qpel_pixels_tab[0][0] = c->avg_h264_qpel_pixels_tab[0][0]; - c->avg_rv40_qpel_pixels_tab[1][0] = c->avg_h264_qpel_pixels_tab[1][0]; - - 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); +#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 FF_SSE2_IDCT_PERM: - for(i=0; i<64; i++) - c->idct_permutation[i]= (i&0x38) | idct_sse2_row_perm[i&7]; + 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_ERROR, "Internal error, IDCT permutation not set\n"); + BIT_DEPTH_FUNCS(8, _16); + break; } -} + + if (HAVE_MMX) ff_dsputil_init_mmx (c, avctx); + if (ARCH_ARM) ff_dsputil_init_arm (c, avctx); + if (HAVE_VIS) ff_dsputil_init_vis (c, avctx); + if (ARCH_ALPHA) ff_dsputil_init_alpha (c, avctx); + if (ARCH_PPC) ff_dsputil_init_ppc (c, avctx); + if (HAVE_MMI) ff_dsputil_init_mmi (c, avctx); + if (ARCH_SH4) ff_dsputil_init_sh4 (c, avctx); + if (ARCH_BFIN) ff_dsputil_init_bfin (c, avctx); + + for (i = 0; i < 4; i++) { + for (j = 0; j < 16; j++) { + if(!c->put_2tap_qpel_pixels_tab[i][j]) + c->put_2tap_qpel_pixels_tab[i][j] = + c->put_h264_qpel_pixels_tab[i][j]; + if(!c->avg_2tap_qpel_pixels_tab[i][j]) + c->avg_2tap_qpel_pixels_tab[i][j] = + c->avg_h264_qpel_pixels_tab[i][j]; + } + } + + ff_init_scantable_permutation(c->idct_permutation, + c->idct_permutation_type); +}