From e7a972e113ddf3271c4c0e01a2d57e23ac5195f1 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Wed, 20 Jul 2011 16:05:05 +0100 Subject: simple_idct: add 10-bit version Signed-off-by: Mans Rullgard --- libavcodec/alpha/dsputil_alpha.c | 2 +- libavcodec/arm/dsputil_init_arm.c | 2 +- libavcodec/arm/dsputil_init_armv5te.c | 5 +- libavcodec/arm/dsputil_init_armv6.c | 5 +- libavcodec/arm/dsputil_init_neon.c | 2 +- libavcodec/bfin/dsputil_bfin.c | 22 +- libavcodec/dct-test.c | 2 +- libavcodec/dsputil.c | 15 +- libavcodec/ppc/dsputil_ppc.c | 2 +- libavcodec/ps2/dsputil_mmi.c | 3 +- libavcodec/sh4/dsputil_sh4.c | 3 +- libavcodec/simple_idct.c | 338 +--------------------------- libavcodec/simple_idct.h | 11 +- libavcodec/simple_idct_template.c | 401 ++++++++++++++++++++++++++++++++++ libavcodec/sparc/dsputil_vis.c | 3 +- libavcodec/vc1.c | 4 +- libavcodec/x86/dsputil_mmx.c | 2 +- 17 files changed, 461 insertions(+), 361 deletions(-) create mode 100644 libavcodec/simple_idct_template.c (limited to 'libavcodec') diff --git a/libavcodec/alpha/dsputil_alpha.c b/libavcodec/alpha/dsputil_alpha.c index 32bb0fc932..039608b9ed 100644 --- a/libavcodec/alpha/dsputil_alpha.c +++ b/libavcodec/alpha/dsputil_alpha.c @@ -335,7 +335,7 @@ void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx) put_pixels_clamped_axp_p = c->put_pixels_clamped; add_pixels_clamped_axp_p = c->add_pixels_clamped; - if (!avctx->lowres && + if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && (avctx->idct_algo == FF_IDCT_AUTO || avctx->idct_algo == FF_IDCT_SIMPLEALPHA)) { c->idct_put = ff_simple_idct_put_axp; diff --git a/libavcodec/arm/dsputil_init_arm.c b/libavcodec/arm/dsputil_init_arm.c index 777a2f954e..5b4b24a43e 100644 --- a/libavcodec/arm/dsputil_init_arm.c +++ b/libavcodec/arm/dsputil_init_arm.c @@ -80,7 +80,7 @@ void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx) ff_put_pixels_clamped = c->put_pixels_clamped; ff_add_pixels_clamped = c->add_pixels_clamped; - if (!avctx->lowres) { + if (!avctx->lowres && avctx->bits_per_raw_sample <= 8) { if(avctx->idct_algo == FF_IDCT_AUTO || avctx->idct_algo == FF_IDCT_ARM){ c->idct_put = j_rev_dct_arm_put; diff --git a/libavcodec/arm/dsputil_init_armv5te.c b/libavcodec/arm/dsputil_init_armv5te.c index 572e06cf36..e0224dabfb 100644 --- a/libavcodec/arm/dsputil_init_armv5te.c +++ b/libavcodec/arm/dsputil_init_armv5te.c @@ -29,8 +29,9 @@ void ff_prefetch_arm(void *mem, int stride, int h); void av_cold ff_dsputil_init_armv5te(DSPContext* c, AVCodecContext *avctx) { - if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO || - avctx->idct_algo == FF_IDCT_SIMPLEARMV5TE)) { + if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && + (avctx->idct_algo == FF_IDCT_AUTO || + avctx->idct_algo == FF_IDCT_SIMPLEARMV5TE)) { c->idct_put = ff_simple_idct_put_armv5te; c->idct_add = ff_simple_idct_add_armv5te; c->idct = ff_simple_idct_armv5te; diff --git a/libavcodec/arm/dsputil_init_armv6.c b/libavcodec/arm/dsputil_init_armv6.c index 7584aeefc6..d442415aec 100644 --- a/libavcodec/arm/dsputil_init_armv6.c +++ b/libavcodec/arm/dsputil_init_armv6.c @@ -74,8 +74,9 @@ void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx) { const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; - if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO || - avctx->idct_algo == FF_IDCT_SIMPLEARMV6)) { + if (!avctx->lowres && avctx->bits_per_raw_sample <= 8 && + (avctx->idct_algo == FF_IDCT_AUTO || + avctx->idct_algo == FF_IDCT_SIMPLEARMV6)) { c->idct_put = ff_simple_idct_put_armv6; c->idct_add = ff_simple_idct_add_armv6; c->idct = ff_simple_idct_armv6; diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c index 3f58dea9cd..ce45caf53e 100644 --- a/libavcodec/arm/dsputil_init_neon.c +++ b/libavcodec/arm/dsputil_init_neon.c @@ -177,7 +177,7 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx) { const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; - if (!avctx->lowres) { + if (!avctx->lowres && avctx->bits_per_raw_sample <= 8) { if (avctx->idct_algo == FF_IDCT_AUTO || avctx->idct_algo == FF_IDCT_SIMPLENEON) { c->idct_put = ff_simple_idct_put_neon; diff --git a/libavcodec/bfin/dsputil_bfin.c b/libavcodec/bfin/dsputil_bfin.c index 0db2d8baf8..d06bd8e4fd 100644 --- a/libavcodec/bfin/dsputil_bfin.c +++ b/libavcodec/bfin/dsputil_bfin.c @@ -256,16 +256,18 @@ void dsputil_init_bfin( DSPContext* c, AVCodecContext *avctx ) if (avctx->dct_algo == FF_DCT_AUTO) c->fdct = ff_bfin_fdct; - if (avctx->idct_algo==FF_IDCT_VP3) { - c->idct_permutation_type = FF_NO_IDCT_PERM; - c->idct = ff_bfin_vp3_idct; - c->idct_add = ff_bfin_vp3_idct_add; - c->idct_put = ff_bfin_vp3_idct_put; - } else if (avctx->idct_algo == FF_IDCT_AUTO) { - c->idct_permutation_type = FF_NO_IDCT_PERM; - c->idct = ff_bfin_idct; - c->idct_add = bfin_idct_add; - c->idct_put = bfin_idct_put; + if (avctx->bits_per_raw_sample <= 8) { + if (avctx->idct_algo == FF_IDCT_VP3) { + c->idct_permutation_type = FF_NO_IDCT_PERM; + c->idct = ff_bfin_vp3_idct; + c->idct_add = ff_bfin_vp3_idct_add; + c->idct_put = ff_bfin_vp3_idct_put; + } else if (avctx->idct_algo == FF_IDCT_AUTO) { + c->idct_permutation_type = FF_NO_IDCT_PERM; + c->idct = ff_bfin_idct; + c->idct_add = bfin_idct_add; + c->idct_put = bfin_idct_put; + } } } diff --git a/libavcodec/dct-test.c b/libavcodec/dct-test.c index dbf9639c1e..136f5c4742 100644 --- a/libavcodec/dct-test.c +++ b/libavcodec/dct-test.c @@ -111,7 +111,7 @@ static const struct algo idct_tab[] = { { "FAANI", ff_faanidct, NO_PERM }, { "REF-DBL", ff_ref_idct, NO_PERM }, { "INT", j_rev_dct, MMX_PERM }, - { "SIMPLE-C", ff_simple_idct, NO_PERM }, + { "SIMPLE-C", ff_simple_idct_8, NO_PERM }, #if HAVE_MMX #if CONFIG_GPL diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index e2cf708940..4b26f68387 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -2237,7 +2237,7 @@ 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]); @@ -2878,6 +2878,12 @@ av_cold 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; @@ -2908,11 +2914,12 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx) 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; diff --git a/libavcodec/ppc/dsputil_ppc.c b/libavcodec/ppc/dsputil_ppc.c index 3bf410c2ca..c1f68fc5f7 100644 --- a/libavcodec/ppc/dsputil_ppc.c +++ b/libavcodec/ppc/dsputil_ppc.c @@ -178,7 +178,7 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx) } #endif //CONFIG_ENCODERS - if (avctx->lowres==0) { + if (avctx->lowres == 0 && avctx->bits_per_raw_sample <= 8) { if ((avctx->idct_algo == FF_IDCT_AUTO) || (avctx->idct_algo == FF_IDCT_ALTIVEC)) { c->idct_put = idct_put_altivec; diff --git a/libavcodec/ps2/dsputil_mmi.c b/libavcodec/ps2/dsputil_mmi.c index f4503a9030..585709679b 100644 --- a/libavcodec/ps2/dsputil_mmi.c +++ b/libavcodec/ps2/dsputil_mmi.c @@ -156,7 +156,8 @@ void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx) c->get_pixels = get_pixels_mmi; - if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2){ + if (avctx->bits_per_raw_sample <= 8 && + (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2)) { c->idct_put= ff_mmi_idct_put; c->idct_add= ff_mmi_idct_add; c->idct = ff_mmi_idct; diff --git a/libavcodec/sh4/dsputil_sh4.c b/libavcodec/sh4/dsputil_sh4.c index 9ea48ad4a1..bf58a9cbbd 100644 --- a/libavcodec/sh4/dsputil_sh4.c +++ b/libavcodec/sh4/dsputil_sh4.c @@ -97,7 +97,8 @@ void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx) if (!high_bit_depth) c->clear_blocks = clear_blocks_sh4; - if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SH4){ + if (avctx->bits_per_raw_sample <= 8 && + (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SH4)) { c->idct_put = idct_put; c->idct_add = idct_add; c->idct = idct_sh4; diff --git a/libavcodec/simple_idct.c b/libavcodec/simple_idct.c index 0ffbfcaf33..b62658bdb3 100644 --- a/libavcodec/simple_idct.c +++ b/libavcodec/simple_idct.c @@ -25,339 +25,19 @@ * simpleidct in C. */ -/* - based upon some outcommented c code from mpeg2dec (idct_mmx.c - written by Aaron Holtzman ) - */ - #include "libavutil/intreadwrite.h" #include "avcodec.h" #include "dsputil.h" #include "mathops.h" #include "simple_idct.h" -#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 - -static inline void idctRowCondDC (DCTELEM * row) -{ - int a0, a1, a2, a3, b0, b1, b2, b3; - -#if HAVE_FAST_64BIT -#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN) - if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) { - uint64_t 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])) { - uint32_t 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]; - - b0 = MUL16(W1, row[1]); - MAC16(b0, W3, row[3]); - b1 = MUL16(W3, row[1]); - MAC16(b1, -W7, row[3]); - b2 = MUL16(W5, row[1]); - MAC16(b2, -W1, row[3]); - b3 = MUL16(W7, row[1]); - MAC16(b3, -W5, row[3]); - - if (AV_RN64A(row + 4)) { - 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; -} - -static inline void idctSparseColPut (uint8_t *dest, int line_size, - DCTELEM * col) -{ - int a0, a1, a2, a3, b0, b1, b2, b3; - uint8_t *cm = ff_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]; - - b0 = MUL16(W1, col[8*1]); - b1 = MUL16(W3, col[8*1]); - b2 = MUL16(W5, col[8*1]); - b3 = MUL16(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]; -} - -static inline void idctSparseColAdd (uint8_t *dest, int line_size, - DCTELEM * col) -{ - int a0, a1, a2, a3, b0, b1, b2, b3; - uint8_t *cm = ff_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]; - - b0 = MUL16(W1, col[8*1]); - b1 = MUL16(W3, col[8*1]); - b2 = MUL16(W5, col[8*1]); - b3 = MUL16(W7, col[8*1]); +#define BIT_DEPTH 8 +#include "simple_idct_template.c" +#undef BIT_DEPTH - 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)]; -} - -static inline void idctSparseCol (DCTELEM * col) -{ - int a0, a1, a2, a3, b0, b1, b2, b3; - - /* 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]; - - b0 = MUL16(W1, col[8*1]); - b1 = MUL16(W3, col[8*1]); - b2 = MUL16(W5, col[8*1]); - b3 = MUL16(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]); - } - - col[0 ] = ((a0 + b0) >> COL_SHIFT); - col[8 ] = ((a1 + b1) >> COL_SHIFT); - col[16] = ((a2 + b2) >> COL_SHIFT); - col[24] = ((a3 + b3) >> COL_SHIFT); - col[32] = ((a3 - b3) >> COL_SHIFT); - col[40] = ((a2 - b2) >> COL_SHIFT); - col[48] = ((a1 - b1) >> COL_SHIFT); - col[56] = ((a0 - b0) >> COL_SHIFT); -} - -void ff_simple_idct_put(uint8_t *dest, int line_size, DCTELEM *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); -} - -void ff_simple_idct_add(uint8_t *dest, int line_size, DCTELEM *block) -{ - 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); -} - -void ff_simple_idct(DCTELEM *block) -{ - int i; - for(i=0; i<8; i++) - idctRowCondDC(block + i*8); - - for(i=0; i<8; i++) - idctSparseCol(block + i); -} +#define BIT_DEPTH 10 +#include "simple_idct_template.c" +#undef BIT_DEPTH /* 2x4x8 idct */ @@ -428,7 +108,7 @@ void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block) /* IDCT8 on each line */ for(i=0; i<8; i++) { - idctRowCondDC(block + i*8); + idctRowCondDC_8(block + i*8); } /* IDCT4 and store */ @@ -503,7 +183,7 @@ void ff_simple_idct84_add(uint8_t *dest, int line_size, DCTELEM *block) /* IDCT8 on each line */ for(i=0; i<4; i++) { - idctRowCondDC(block + i*8); + idctRowCondDC_8(block + i*8); } /* IDCT4 and store */ @@ -523,7 +203,7 @@ void ff_simple_idct48_add(uint8_t *dest, int line_size, DCTELEM *block) /* IDCT8 and store */ for(i=0; i<4; i++){ - idctSparseColAdd(dest + i, line_size, block + i); + idctSparseColAdd_8(dest + i, line_size, block + i); } } diff --git a/libavcodec/simple_idct.h b/libavcodec/simple_idct.h index 23bae9c2fe..a33eb964ce 100644 --- a/libavcodec/simple_idct.h +++ b/libavcodec/simple_idct.h @@ -31,12 +31,17 @@ #include #include "dsputil.h" -void ff_simple_idct_put(uint8_t *dest, int line_size, DCTELEM *block); -void ff_simple_idct_add(uint8_t *dest, int line_size, DCTELEM *block); +void ff_simple_idct_put_8(uint8_t *dest, int line_size, DCTELEM *block); +void ff_simple_idct_add_8(uint8_t *dest, int line_size, DCTELEM *block); +void ff_simple_idct_8(DCTELEM *block); + +void ff_simple_idct_put_10(uint8_t *dest, int line_size, DCTELEM *block); +void ff_simple_idct_add_10(uint8_t *dest, int line_size, DCTELEM *block); +void ff_simple_idct_10(DCTELEM *block); + void ff_simple_idct_mmx(int16_t *block); void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block); void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block); -void ff_simple_idct(DCTELEM *block); void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block); diff --git a/libavcodec/simple_idct_template.c b/libavcodec/simple_idct_template.c new file mode 100644 index 0000000000..be49cb9570 --- /dev/null +++ b/libavcodec/simple_idct_template.c @@ -0,0 +1,401 @@ +/* + * Simple IDCT + * + * Copyright (c) 2001 Michael Niedermayer + * + * 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 2.1 of the License, or (at your option) any later version. + * + * 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 Libav; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * simpleidct in C. + */ + +/* + based upon some outcommented c code from mpeg2dec (idct_mmx.c + written by Aaron Holtzman ) + */ + +#include "bit_depth_template.c" + +#undef W1 +#undef W2 +#undef W3 +#undef W4 +#undef W5 +#undef W6 +#undef W7 +#undef ROW_SHIFT +#undef COL_SHIFT +#undef DC_SHIFT +#undef MUL +#undef MAC + +#if BIT_DEPTH == 8 + +#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 +#define DC_SHIFT 3 + +#define MUL(a, b) MUL16(a, b) +#define MAC(a, b, c) MAC16(a, b, c) + +#elif BIT_DEPTH == 10 + +#define W1 90901 +#define W2 85627 +#define W3 77062 +#define W4 65535 +#define W5 51491 +#define W6 35468 +#define W7 18081 + +#define ROW_SHIFT 15 +#define COL_SHIFT 20 +#define DC_SHIFT 1 + +#define MUL(a, b) ((a) * (b)) +#define MAC(a, b, c) ((a) += (b) * (c)) + +#else + +#error "Unsupported bitdepth" + +#endif + +static inline void FUNC(idctRowCondDC)(DCTELEM *row) +{ + int a0, a1, a2, a3, b0, b1, b2, b3; + +#if HAVE_FAST_64BIT +#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN) + if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) { + uint64_t temp = (row[0] << DC_SHIFT) & 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])) { + uint32_t temp = (row[0] << DC_SHIFT) & 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]; + + b0 = MUL(W1, row[1]); + MAC(b0, W3, row[3]); + b1 = MUL(W3, row[1]); + MAC(b1, -W7, row[3]); + b2 = MUL(W5, row[1]); + MAC(b2, -W1, row[3]); + b3 = MUL(W7, row[1]); + MAC(b3, -W5, row[3]); + + if (AV_RN64A(row + 4)) { + 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]; + + MAC(b0, W5, row[5]); + MAC(b0, W7, row[7]); + + MAC(b1, -W1, row[5]); + MAC(b1, -W5, row[7]); + + MAC(b2, W7, row[5]); + MAC(b2, W3, row[7]); + + MAC(b3, W3, row[5]); + MAC(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; +} + +static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size, + DCTELEM *col) +{ + int a0, a1, a2, a3, b0, b1, b2, b3; + INIT_CLIP; + + /* 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]; + + b0 = MUL(W1, col[8*1]); + b1 = MUL(W3, col[8*1]); + b2 = MUL(W5, col[8*1]); + b3 = MUL(W7, col[8*1]); + + MAC(b0, + W3, col[8*3]); + MAC(b1, - W7, col[8*3]); + MAC(b2, - W1, col[8*3]); + MAC(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]) { + MAC(b0, + W5, col[8*5]); + MAC(b1, - W1, col[8*5]); + MAC(b2, + W7, col[8*5]); + MAC(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]) { + MAC(b0, + W7, col[8*7]); + MAC(b1, - W5, col[8*7]); + MAC(b2, + W3, col[8*7]); + MAC(b3, - W1, col[8*7]); + } + + dest[0] = CLIP((a0 + b0) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a1 + b1) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a2 + b2) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a3 + b3) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a3 - b3) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a2 - b2) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a1 - b1) >> COL_SHIFT); + dest += line_size; + dest[0] = CLIP((a0 - b0) >> COL_SHIFT); +} + +static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size, + DCTELEM *col) +{ + int a0, a1, a2, a3, b0, b1, b2, b3; + INIT_CLIP; + + /* 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]; + + b0 = MUL(W1, col[8*1]); + b1 = MUL(W3, col[8*1]); + b2 = MUL(W5, col[8*1]); + b3 = MUL(W7, col[8*1]); + + MAC(b0, + W3, col[8*3]); + MAC(b1, - W7, col[8*3]); + MAC(b2, - W1, col[8*3]); + MAC(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]) { + MAC(b0, + W5, col[8*5]); + MAC(b1, - W1, col[8*5]); + MAC(b2, + W7, col[8*5]); + MAC(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]) { + MAC(b0, + W7, col[8*7]); + MAC(b1, - W5, col[8*7]); + MAC(b2, + W3, col[8*7]); + MAC(b3, - W1, col[8*7]); + } + + dest[0] = CLIP(dest[0] + ((a0 + b0) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a1 + b1) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a2 + b2) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a3 + b3) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a3 - b3) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a2 - b2) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a1 - b1) >> COL_SHIFT)); + dest += line_size; + dest[0] = CLIP(dest[0] + ((a0 - b0) >> COL_SHIFT)); +} + +static inline void FUNC(idctSparseCol)(DCTELEM *col) +{ + int a0, a1, a2, a3, b0, b1, b2, b3; + + /* 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]; + + b0 = MUL(W1, col[8*1]); + b1 = MUL(W3, col[8*1]); + b2 = MUL(W5, col[8*1]); + b3 = MUL(W7, col[8*1]); + + MAC(b0, + W3, col[8*3]); + MAC(b1, - W7, col[8*3]); + MAC(b2, - W1, col[8*3]); + MAC(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]) { + MAC(b0, + W5, col[8*5]); + MAC(b1, - W1, col[8*5]); + MAC(b2, + W7, col[8*5]); + MAC(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]) { + MAC(b0, + W7, col[8*7]); + MAC(b1, - W5, col[8*7]); + MAC(b2, + W3, col[8*7]); + MAC(b3, - W1, col[8*7]); + } + + col[0 ] = ((a0 + b0) >> COL_SHIFT); + col[8 ] = ((a1 + b1) >> COL_SHIFT); + col[16] = ((a2 + b2) >> COL_SHIFT); + col[24] = ((a3 + b3) >> COL_SHIFT); + col[32] = ((a3 - b3) >> COL_SHIFT); + col[40] = ((a2 - b2) >> COL_SHIFT); + col[48] = ((a1 - b1) >> COL_SHIFT); + col[56] = ((a0 - b0) >> COL_SHIFT); +} + +void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, DCTELEM *block) +{ + pixel *dest = (pixel *)dest_; + int i; + for(i=0; i<8; i++) + FUNC(idctRowCondDC)(block + i*8); + + for(i=0; i<8; i++) + FUNC(idctSparseColPut)(dest + i, line_size, block + i); +} + +void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, DCTELEM *block) +{ + pixel *dest = (pixel *)dest_; + int i; + for(i=0; i<8; i++) + FUNC(idctRowCondDC)(block + i*8); + + for(i=0; i<8; i++) + FUNC(idctSparseColAdd)(dest + i, line_size, block + i); +} + +void FUNC(ff_simple_idct)(DCTELEM *block) +{ + int i; + for(i=0; i<8; i++) + FUNC(idctRowCondDC)(block + i*8); + + for(i=0; i<8; i++) + FUNC(idctSparseCol)(block + i); +} diff --git a/libavcodec/sparc/dsputil_vis.c b/libavcodec/sparc/dsputil_vis.c index ab9258b2b9..9811b2622e 100644 --- a/libavcodec/sparc/dsputil_vis.c +++ b/libavcodec/sparc/dsputil_vis.c @@ -3956,7 +3956,8 @@ void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx) const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8; if (accel & ACCEL_SPARC_VIS) { - if(avctx->idct_algo==FF_IDCT_SIMPLEVIS){ + if (avctx->bits_per_raw_sample <= 8 && + avctx->idct_algo == FF_IDCT_SIMPLEVIS) { c->idct_put = ff_simple_idct_put_vis; c->idct_add = ff_simple_idct_add_vis; c->idct = ff_simple_idct_vis; diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c index 32869b97d1..1f83dd27b8 100644 --- a/libavcodec/vc1.c +++ b/libavcodec/vc1.c @@ -337,11 +337,11 @@ int vc1_decode_sequence_header(AVCodecContext *avctx, VC1Context *v, GetBitConte v->res_fasttx = get_bits1(gb); if (!v->res_fasttx) { - v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct; + v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct_8; v->vc1dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add; v->vc1dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add; v->vc1dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add; - v->vc1dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add; + v->vc1dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add_8; v->vc1dsp.vc1_inv_trans_8x4_dc = ff_simple_idct84_add; v->vc1dsp.vc1_inv_trans_4x8_dc = ff_simple_idct48_add; v->vc1dsp.vc1_inv_trans_4x4_dc = ff_simple_idct44_add; diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c index 29fff276ad..84ae47b04d 100644 --- a/libavcodec/x86/dsputil_mmx.c +++ b/libavcodec/x86/dsputil_mmx.c @@ -2465,7 +2465,7 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx) if (mm_flags & AV_CPU_FLAG_MMX) { const int idct_algo= avctx->idct_algo; - if(avctx->lowres==0){ + if (avctx->lowres == 0 && avctx->bits_per_raw_sample <= 8) { if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SIMPLEMMX){ c->idct_put= ff_simple_idct_put_mmx; c->idct_add= ff_simple_idct_add_mmx; -- cgit v1.2.3