X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fsimple_idct.c;h=6ee1320394bc53d2bc7c0960b3a788fa89a31f98;hb=1f1ad8ace040a08edc2646ff638ca42a8828779f;hp=0665f667afd79858d7cbf8e9a24ed8f5998b72bf;hpb=d36a2466753fc18f3143837a56aa7123f8991f33;p=ffmpeg diff --git a/libavcodec/simple_idct.c b/libavcodec/simple_idct.c index 0665f667afd..6ee1320394b 100644 --- a/libavcodec/simple_idct.c +++ b/libavcodec/simple_idct.c @@ -3,508 +3,233 @@ * * Copyright (c) 2001 Michael Niedermayer * - * This library is free software; you can redistribute it and/or + * This file is part of Libav. + * + * 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 of the License, or (at your option) any later version. + * version 2.1 of the License, or (at your option) any later version. * - * This library 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 this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * License along with Libav; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ -/* - based upon some outcommented c code from mpeg2dec (idct_mmx.c - written by Aaron Holtzman ) + +/** + * @file + * simpleidct in C. */ + +#include "libavutil/intreadwrite.h" #include "avcodec.h" -#include "dsputil.h" +#include "mathops.h" #include "simple_idct.h" -#if 0 -#define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ -#define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */ -#define W3 2408 /* 2048*sqrt (2)*cos (3*pi/16) */ -#define W4 2048 /* 2048*sqrt (2)*cos (4*pi/16) */ -#define W5 1609 /* 2048*sqrt (2)*cos (5*pi/16) */ -#define W6 1108 /* 2048*sqrt (2)*cos (6*pi/16) */ -#define W7 565 /* 2048*sqrt (2)*cos (7*pi/16) */ -#define ROW_SHIFT 8 -#define COL_SHIFT 17 -#else -#define W1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define W7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 -#define ROW_SHIFT 11 -#define COL_SHIFT 20 // 6 -#endif - -#ifdef ARCH_ALPHA -#define FAST_64BIT -#endif - -#if defined(ARCH_POWERPC_405) - -/* signed 16x16 -> 32 multiply add accumulate */ -#define MAC16(rt, ra, rb) \ - asm ("maclhw %0, %2, %3" : "=r" (rt) : "0" (rt), "r" (ra), "r" (rb)); - -/* signed 16x16 -> 32 multiply */ -#define MUL16(rt, ra, rb) \ - asm ("mullhw %0, %1, %2" : "=r" (rt) : "r" (ra), "r" (rb)); +#define BIT_DEPTH 8 +#include "simple_idct_template.c" +#undef BIT_DEPTH -#else +#define BIT_DEPTH 10 +#include "simple_idct_template.c" +#undef BIT_DEPTH -/* signed 16x16 -> 32 multiply add accumulate */ -#define MAC16(rt, ra, rb) rt += (ra) * (rb) +/* 2x4x8 idct */ -/* signed 16x16 -> 32 multiply */ -#define MUL16(rt, ra, rb) rt = (ra) * (rb) +#define CN_SHIFT 12 +#define C_FIX(x) ((int)((x) * (1 << CN_SHIFT) + 0.5)) +#define C1 C_FIX(0.6532814824) +#define C2 C_FIX(0.2705980501) -#endif +/* row idct is multiple by 16 * sqrt(2.0), col idct4 is normalized, + and the butterfly must be multiplied by 0.5 * sqrt(2.0) */ +#define C_SHIFT (4+1+12) -#ifdef ARCH_ALPHA -/* 0: all entries 0, 1: only first entry nonzero, 2: otherwise */ -static inline int idctRowCondDC(int16_t *row) +static inline void idct4col_put(uint8_t *dest, int line_size, const int16_t *col) { - int_fast32_t a0, a1, a2, a3, b0, b1, b2, b3; - uint64_t *lrow = (uint64_t *) row; - - if (lrow[1] == 0) { - if (lrow[0] == 0) - return 0; - if ((lrow[0] & ~0xffffULL) == 0) { - uint64_t v; - - a0 = W4 * row[0]; - a0 += 1 << (ROW_SHIFT - 1); - a0 >>= ROW_SHIFT; - v = (uint16_t) a0; - v += v << 16; - v += v << 32; - lrow[0] = v; - lrow[1] = v; - - return 1; - } - } - - a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); - a1 = a0; - a2 = a0; - a3 = a0; - - if (row[2]) { - a0 += W2 * row[2]; - a1 += W6 * row[2]; - a2 -= W6 * row[2]; - a3 -= W2 * row[2]; - } - - if (row[4]) { - a0 += W4 * row[4]; - a1 -= W4 * row[4]; - a2 -= W4 * row[4]; - a3 += W4 * row[4]; - } - - if (row[6]) { - a0 += W6 * row[6]; - a1 -= W2 * row[6]; - a2 += W2 * row[6]; - a3 -= W6 * row[6]; - } - - if (row[1]) { - b0 = W1 * row[1]; - b1 = W3 * row[1]; - b2 = W5 * row[1]; - b3 = W7 * row[1]; - } else { - b0 = 0; - b1 = 0; - b2 = 0; - b3 = 0; - } - - if (row[3]) { - b0 += W3 * row[3]; - b1 -= W7 * row[3]; - b2 -= W1 * row[3]; - b3 -= W5 * row[3]; - } - - if (row[5]) { - b0 += W5 * row[5]; - b1 -= W1 * row[5]; - b2 += W7 * row[5]; - b3 += W3 * row[5]; - } - - if (row[7]) { - b0 += W7 * row[7]; - b1 -= W5 * row[7]; - b2 += W3 * row[7]; - b3 -= W1 * row[7]; - } - - row[0] = (a0 + b0) >> ROW_SHIFT; - row[1] = (a1 + b1) >> ROW_SHIFT; - row[2] = (a2 + b2) >> ROW_SHIFT; - row[3] = (a3 + b3) >> ROW_SHIFT; - row[4] = (a3 - b3) >> ROW_SHIFT; - row[5] = (a2 - b2) >> ROW_SHIFT; - row[6] = (a1 - b1) >> ROW_SHIFT; - row[7] = (a0 - b0) >> ROW_SHIFT; + int c0, c1, c2, c3, a0, a1, a2, a3; + + a0 = col[8*0]; + a1 = col[8*2]; + a2 = col[8*4]; + a3 = col[8*6]; + c0 = ((a0 + a2) << (CN_SHIFT - 1)) + (1 << (C_SHIFT - 1)); + c2 = ((a0 - a2) << (CN_SHIFT - 1)) + (1 << (C_SHIFT - 1)); + c1 = a1 * C1 + a3 * C2; + c3 = a1 * C2 - a3 * C1; + dest[0] = av_clip_uint8((c0 + c1) >> C_SHIFT); + dest += line_size; + dest[0] = av_clip_uint8((c2 + c3) >> C_SHIFT); + dest += line_size; + dest[0] = av_clip_uint8((c2 - c3) >> C_SHIFT); + dest += line_size; + dest[0] = av_clip_uint8((c0 - c1) >> C_SHIFT); +} - return 2; +#define BF(k) \ +{\ + int a0, a1;\ + a0 = ptr[k];\ + a1 = ptr[8 + k];\ + ptr[k] = a0 + a1;\ + ptr[8 + k] = a0 - a1;\ } -#else /* not ARCH_ALPHA */ -static inline void idctRowCondDC (int16_t * row) +/* only used by DV codec. The input must be interlaced. 128 is added + to the pixels before clamping to avoid systematic error + (1024*sqrt(2)) offset would be needed otherwise. */ +/* XXX: I think a 1.0/sqrt(2) normalization should be needed to + compensate the extra butterfly stage - I don't have the full DV + specification */ +void ff_simple_idct248_put(uint8_t *dest, int line_size, int16_t *block) { - int a0, a1, a2, a3, b0, b1, b2, b3; -#ifdef FAST_64BIT - uint64_t temp; -#else - uint32_t temp; -#endif - -#ifdef FAST_64BIT -#ifdef WORDS_BIGENDIAN -#define ROW0_MASK 0xffff000000000000LL -#else -#define ROW0_MASK 0xffffLL -#endif - if ( ((((uint64_t *)row)[0] & ~ROW0_MASK) | - ((uint64_t *)row)[1]) == 0) { - temp = (row[0] << 3) & 0xffff; - temp += temp << 16; - temp += temp << 32; - ((uint64_t *)row)[0] = temp; - ((uint64_t *)row)[1] = temp; - return; - } -#else - if (!(((uint32_t*)row)[1] | - ((uint32_t*)row)[2] | - ((uint32_t*)row)[3] | - row[1])) { - temp = (row[0] << 3) & 0xffff; - temp += temp << 16; - ((uint32_t*)row)[0]=((uint32_t*)row)[1] = - ((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp; - return; - } -#endif - - a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); - a1 = a0; - a2 = a0; - a3 = a0; - - /* no need to optimize : gcc does it */ - a0 += W2 * row[2]; - a1 += W6 * row[2]; - a2 -= W6 * row[2]; - a3 -= W2 * row[2]; - - MUL16(b0, W1, row[1]); - MAC16(b0, W3, row[3]); - MUL16(b1, W3, row[1]); - MAC16(b1, -W7, row[3]); - MUL16(b2, W5, row[1]); - MAC16(b2, -W1, row[3]); - MUL16(b3, W7, row[1]); - MAC16(b3, -W5, row[3]); - -#ifdef FAST_64BIT - temp = ((uint64_t*)row)[1]; -#else - temp = ((uint32_t*)row)[2] | ((uint32_t*)row)[3]; -#endif - if (temp != 0) { - a0 += W4*row[4] + W6*row[6]; - a1 += - W4*row[4] - W2*row[6]; - a2 += - W4*row[4] + W2*row[6]; - a3 += W4*row[4] - W6*row[6]; - - MAC16(b0, W5, row[5]); - MAC16(b0, W7, row[7]); - - MAC16(b1, -W1, row[5]); - MAC16(b1, -W5, row[7]); - - MAC16(b2, W7, row[5]); - MAC16(b2, W3, row[7]); - - MAC16(b3, W3, row[5]); - MAC16(b3, -W1, row[7]); - } - - row[0] = (a0 + b0) >> ROW_SHIFT; - row[7] = (a0 - b0) >> ROW_SHIFT; - row[1] = (a1 + b1) >> ROW_SHIFT; - row[6] = (a1 - b1) >> ROW_SHIFT; - row[2] = (a2 + b2) >> ROW_SHIFT; - row[5] = (a2 - b2) >> ROW_SHIFT; - row[3] = (a3 + b3) >> ROW_SHIFT; - row[4] = (a3 - b3) >> ROW_SHIFT; + int i; + int16_t *ptr; + + /* butterfly */ + ptr = block; + for(i=0;i<4;i++) { + BF(0); + BF(1); + BF(2); + BF(3); + BF(4); + BF(5); + BF(6); + BF(7); + ptr += 2 * 8; + } + + /* IDCT8 on each line */ + for(i=0; i<8; i++) { + idctRowCondDC_8(block + i*8, 0); + } + + /* IDCT4 and store */ + for(i=0;i<8;i++) { + idct4col_put(dest + i, 2 * line_size, block + i); + idct4col_put(dest + line_size + i, 2 * line_size, block + 8 + i); + } } -#endif /* not ARCH_ALPHA */ -static inline void idctSparseColPut (UINT8 *dest, int line_size, - int16_t * col) +/* 8x4 & 4x8 WMV2 IDCT */ +#undef CN_SHIFT +#undef C_SHIFT +#undef C_FIX +#undef C1 +#undef C2 +#define CN_SHIFT 12 +#define C_FIX(x) ((int)((x) * 1.414213562 * (1 << CN_SHIFT) + 0.5)) +#define C1 C_FIX(0.6532814824) +#define C2 C_FIX(0.2705980501) +#define C3 C_FIX(0.5) +#define C_SHIFT (4+1+12) +static inline void idct4col_add(uint8_t *dest, int line_size, const int16_t *col) { - int a0, a1, a2, a3, b0, b1, b2, b3; - UINT8 *cm = cropTbl + MAX_NEG_CROP; - - /* XXX: I did that only to give same values as previous code */ - a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); - a1 = a0; - a2 = a0; - a3 = a0; - - a0 += + W2*col[8*2]; - a1 += + W6*col[8*2]; - a2 += - W6*col[8*2]; - a3 += - W2*col[8*2]; - - MUL16(b0, W1, col[8*1]); - MUL16(b1, W3, col[8*1]); - MUL16(b2, W5, col[8*1]); - MUL16(b3, W7, col[8*1]); - - MAC16(b0, + W3, col[8*3]); - MAC16(b1, - W7, col[8*3]); - MAC16(b2, - W1, col[8*3]); - MAC16(b3, - W5, col[8*3]); - - if(col[8*4]){ - a0 += + W4*col[8*4]; - a1 += - W4*col[8*4]; - a2 += - W4*col[8*4]; - a3 += + W4*col[8*4]; - } - - if (col[8*5]) { - MAC16(b0, + W5, col[8*5]); - MAC16(b1, - W1, col[8*5]); - MAC16(b2, + W7, col[8*5]); - MAC16(b3, + W3, col[8*5]); - } - - if(col[8*6]){ - a0 += + W6*col[8*6]; - a1 += - W2*col[8*6]; - a2 += + W2*col[8*6]; - a3 += - W6*col[8*6]; - } - - if (col[8*7]) { - MAC16(b0, + W7, col[8*7]); - MAC16(b1, - W5, col[8*7]); - MAC16(b2, + W3, col[8*7]); - MAC16(b3, - W1, col[8*7]); - } - - dest[0] = cm[(a0 + b0) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a1 + b1) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a2 + b2) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a3 + b3) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a3 - b3) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a2 - b2) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a1 - b1) >> COL_SHIFT]; - dest += line_size; - dest[0] = cm[(a0 - b0) >> COL_SHIFT]; + int c0, c1, c2, c3, a0, a1, a2, a3; + + a0 = col[8*0]; + a1 = col[8*1]; + a2 = col[8*2]; + a3 = col[8*3]; + c0 = (a0 + a2)*C3 + (1 << (C_SHIFT - 1)); + c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1)); + c1 = a1 * C1 + a3 * C2; + c3 = a1 * C2 - a3 * C1; + dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT)); + dest += line_size; + dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT)); + dest += line_size; + dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT)); + dest += line_size; + dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT)); } -static inline void idctSparseColAdd (UINT8 *dest, int line_size, - int16_t * col) +#define RN_SHIFT 15 +#define R_FIX(x) ((int)((x) * 1.414213562 * (1 << RN_SHIFT) + 0.5)) +#define R1 R_FIX(0.6532814824) +#define R2 R_FIX(0.2705980501) +#define R3 R_FIX(0.5) +#define R_SHIFT 11 +static inline void idct4row(int16_t *row) { - int a0, a1, a2, a3, b0, b1, b2, b3; - UINT8 *cm = cropTbl + MAX_NEG_CROP; - - /* XXX: I did that only to give same values as previous code */ - a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); - a1 = a0; - a2 = a0; - a3 = a0; - - a0 += + W2*col[8*2]; - a1 += + W6*col[8*2]; - a2 += - W6*col[8*2]; - a3 += - W2*col[8*2]; - - MUL16(b0, W1, col[8*1]); - MUL16(b1, W3, col[8*1]); - MUL16(b2, W5, col[8*1]); - MUL16(b3, W7, col[8*1]); - - MAC16(b0, + W3, col[8*3]); - MAC16(b1, - W7, col[8*3]); - MAC16(b2, - W1, col[8*3]); - MAC16(b3, - W5, col[8*3]); - - if(col[8*4]){ - a0 += + W4*col[8*4]; - a1 += - W4*col[8*4]; - a2 += - W4*col[8*4]; - a3 += + W4*col[8*4]; - } - - if (col[8*5]) { - MAC16(b0, + W5, col[8*5]); - MAC16(b1, - W1, col[8*5]); - MAC16(b2, + W7, col[8*5]); - MAC16(b3, + W3, col[8*5]); - } - - if(col[8*6]){ - a0 += + W6*col[8*6]; - a1 += - W2*col[8*6]; - a2 += + W2*col[8*6]; - a3 += - W6*col[8*6]; - } - - if (col[8*7]) { - MAC16(b0, + W7, col[8*7]); - MAC16(b1, - W5, col[8*7]); - MAC16(b2, + W3, col[8*7]); - MAC16(b3, - W1, col[8*7]); - } - - dest[0] = cm[dest[0] + ((a0 + b0) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a1 + b1) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a2 + b2) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a3 + b3) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a3 - b3) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a2 - b2) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a1 - b1) >> COL_SHIFT)]; - dest += line_size; - dest[0] = cm[dest[0] + ((a0 - b0) >> COL_SHIFT)]; + int c0, c1, c2, c3, a0, a1, a2, a3; + + a0 = row[0]; + a1 = row[1]; + a2 = row[2]; + a3 = row[3]; + c0 = (a0 + a2)*R3 + (1 << (R_SHIFT - 1)); + c2 = (a0 - a2)*R3 + (1 << (R_SHIFT - 1)); + c1 = a1 * R1 + a3 * R2; + c3 = a1 * R2 - a3 * R1; + row[0]= (c0 + c1) >> R_SHIFT; + row[1]= (c2 + c3) >> R_SHIFT; + row[2]= (c2 - c3) >> R_SHIFT; + row[3]= (c0 - c1) >> R_SHIFT; } -#ifdef ARCH_ALPHA -/* If all rows but the first one are zero after row transformation, - all rows will be identical after column transformation. */ -static inline void idctCol2(int16_t *col) +void ff_simple_idct84_add(uint8_t *dest, int line_size, int16_t *block) { - int i; - uint64_t l, r; - uint64_t *lcol = (uint64_t *) col; - - for (i = 0; i < 8; ++i) { - int a0 = col[0] + (1 << (COL_SHIFT - 1)) / W4; + int i; - a0 *= W4; - col[0] = a0 >> COL_SHIFT; - ++col; - } + /* IDCT8 on each line */ + for(i=0; i<4; i++) { + idctRowCondDC_8(block + i*8, 0); + } - l = lcol[0]; - r = lcol[1]; - lcol[ 2] = l; lcol[ 3] = r; - lcol[ 4] = l; lcol[ 5] = r; - lcol[ 6] = l; lcol[ 7] = r; - lcol[ 8] = l; lcol[ 9] = r; - lcol[10] = l; lcol[11] = r; - lcol[12] = l; lcol[13] = r; - lcol[14] = l; lcol[15] = r; + /* IDCT4 and store */ + for(i=0;i<8;i++) { + idct4col_add(dest + i, line_size, block + i); + } } -void simple_idct (short *block) +void ff_simple_idct48_add(uint8_t *dest, int line_size, int16_t *block) { + int i; - int i; - int rowsZero = 1; /* all rows except row 0 zero */ - int rowsConstant = 1; /* all rows consist of a constant value */ - - for (i = 0; i < 8; i++) { - int sparseness = idctRowCondDC(block + 8 * i); - - if (i > 0 && sparseness > 0) - rowsZero = 0; - if (sparseness == 2) - rowsConstant = 0; - } - - if (rowsZero) { - idctCol2(block); - } else if (rowsConstant) { - uint64_t *lblock = (uint64_t *) block; - - idctSparseCol(block); - for (i = 0; i < 8; i++) { - uint64_t v = (uint16_t) block[i * 8]; - - v += v << 16; - v += v << 32; - lblock[0] = v; - lblock[1] = v; - lblock += 2; - } - } else { - for (i = 0; i < 8; i++) - idctSparseCol(block + i); - } -} + /* IDCT4 on each line */ + for(i=0; i<8; i++) { + idct4row(block + i*8); + } -/* XXX: suppress this mess */ -void simple_idct_put(UINT8 *dest, int line_size, DCTELEM *block) -{ - simple_idct(block); - put_pixels_clamped(block, dest, line_size); + /* IDCT8 and store */ + for(i=0; i<4; i++){ + idctSparseColAdd_8(dest + i, line_size, block + i); + } } -void simple_idct_add(UINT8 *dest, int line_size, DCTELEM *block) +void ff_simple_idct44_add(uint8_t *dest, int line_size, int16_t *block) { - simple_idct(block); - add_pixels_clamped(block, dest, line_size); -} + int i; -#else + /* IDCT4 on each line */ + for(i=0; i<4; i++) { + idct4row(block + i*8); + } -void simple_idct_put(UINT8 *dest, int line_size, INT16 *block) -{ - int i; - for(i=0; i<8; i++) - idctRowCondDC(block + i*8); - - for(i=0; i<8; i++) - idctSparseColPut(dest + i, line_size, block + i); + /* IDCT4 and store */ + for(i=0; i<4; i++){ + idct4col_add(dest + i, line_size, block + i); + } } -void simple_idct_add(UINT8 *dest, int line_size, INT16 *block) +#if CONFIG_PRORES_DECODER +void ff_prores_idct(int16_t *block, const int16_t *qmat) { int i; - for(i=0; i<8; i++) - idctRowCondDC(block + i*8); - - for(i=0; i<8; i++) - idctSparseColAdd(dest + i, line_size, block + i); -} -#endif + for (i = 0; i < 64; i++) + block[i] *= qmat[i]; -#undef COL_SHIFT + for (i = 0; i < 8; i++) + idctRowCondDC_10(block + i*8, 2); + + for (i = 0; i < 8; i++) + idctSparseCol_10(block + i); +} +#endif /* CONFIG_PRORES_DECODER */