X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdct.c;h=0128c7d3e56e939f7f5b910458c0a2ab394f8dee;hb=42b9150b0d4f0a130c1d93dc991fd5412743a8cf;hp=0840feca0abc1d683ae5a412d00771d1c0e8829b;hpb=ba87f0801d77c21eb1e4891ca1f846500bbb0939;p=ffmpeg diff --git a/libavcodec/dct.c b/libavcodec/dct.c index 0840feca0ab..0128c7d3e56 100644 --- a/libavcodec/dct.c +++ b/libavcodec/dct.c @@ -4,20 +4,20 @@ * Copyright (c) 2010 Alex Converse * Copyright (c) 2010 Vitor Sessak * - * 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 St, Fifth Floor, Boston, MA 02110-1301 USA */ @@ -28,14 +28,17 @@ */ #include +#include + #include "libavutil/mathematics.h" -#include "fft.h" +#include "dct.h" +#include "dct32.h" -/* sin((M_PI * x / (2*n)) */ -#define SIN(s,n,x) (s->costab[(n) - (x)]) +/* sin((M_PI * x / (2 * n)) */ +#define SIN(s, n, x) (s->costab[(n) - (x)]) -/* cos((M_PI * x / (2*n)) */ -#define COS(s,n,x) (s->costab[x]) +/* cos((M_PI * x / (2 * n)) */ +#define COS(s, n, x) (s->costab[x]) static void ff_dst_calc_I_c(DCTContext *ctx, FFTSample *data) { @@ -43,28 +46,28 @@ static void ff_dst_calc_I_c(DCTContext *ctx, FFTSample *data) int i; data[0] = 0; - for(i = 1; i < n/2; i++) { - float tmp1 = data[i ]; - float tmp2 = data[n - i]; - float s = SIN(ctx, n, 2*i); - - s *= tmp1 + tmp2; - tmp1 = (tmp1 - tmp2) * 0.5f; - data[i ] = s + tmp1; - data[n - i] = s - tmp1; + for (i = 1; i < n / 2; i++) { + float tmp1 = data[i ]; + float tmp2 = data[n - i]; + float s = SIN(ctx, n, 2 * i); + + s *= tmp1 + tmp2; + tmp1 = (tmp1 - tmp2) * 0.5f; + data[i] = s + tmp1; + data[n - i] = s - tmp1; } - data[n/2] *= 2; - ff_rdft_calc(&ctx->rdft, data); + data[n / 2] *= 2; + ctx->rdft.rdft_calc(&ctx->rdft, data); data[0] *= 0.5f; - for(i = 1; i < n-2; i += 2) { - data[i + 1] += data[i - 1]; - data[i ] = -data[i + 2]; + for (i = 1; i < n - 2; i += 2) { + data[i + 1] += data[i - 1]; + data[i] = -data[i + 2]; } - data[n-1] = 0; + data[n - 1] = 0; } static void ff_dct_calc_I_c(DCTContext *ctx, FFTSample *data) @@ -73,27 +76,27 @@ static void ff_dct_calc_I_c(DCTContext *ctx, FFTSample *data) int i; float next = -0.5f * (data[0] - data[n]); - for(i = 0; i < n/2; i++) { - float tmp1 = data[i ]; + for (i = 0; i < n / 2; i++) { + float tmp1 = data[i]; float tmp2 = data[n - i]; - float s = SIN(ctx, n, 2*i); - float c = COS(ctx, n, 2*i); + float s = SIN(ctx, n, 2 * i); + float c = COS(ctx, n, 2 * i); c *= tmp1 - tmp2; s *= tmp1 - tmp2; next += c; - tmp1 = (tmp1 + tmp2) * 0.5f; - data[i ] = tmp1 - s; + tmp1 = (tmp1 + tmp2) * 0.5f; + data[i] = tmp1 - s; data[n - i] = tmp1 + s; } - ff_rdft_calc(&ctx->rdft, data); + ctx->rdft.rdft_calc(&ctx->rdft, data); data[n] = data[1]; data[1] = next; - for(i = 3; i <= n; i += 2) + for (i = 3; i <= n; i += 2) data[i] = data[i - 2] - data[i]; } @@ -102,31 +105,31 @@ static void ff_dct_calc_III_c(DCTContext *ctx, FFTSample *data) int n = 1 << ctx->nbits; int i; - float next = data[n - 1]; + float next = data[n - 1]; float inv_n = 1.0f / n; for (i = n - 2; i >= 2; i -= 2) { - float val1 = data[i ]; + float val1 = data[i]; float val2 = data[i - 1] - data[i + 1]; - float c = COS(ctx, n, i); - float s = SIN(ctx, n, i); + float c = COS(ctx, n, i); + float s = SIN(ctx, n, i); - data[i ] = c * val1 + s * val2; + data[i] = c * val1 + s * val2; data[i + 1] = s * val1 - c * val2; } data[1] = 2 * next; - ff_rdft_calc(&ctx->rdft, data); + ctx->rdft.rdft_calc(&ctx->rdft, data); for (i = 0; i < n / 2; i++) { - float tmp1 = data[i ] * inv_n; + float tmp1 = data[i] * inv_n; float tmp2 = data[n - i - 1] * inv_n; - float csc = ctx->csc2[i] * (tmp1 - tmp2); + float csc = ctx->csc2[i] * (tmp1 - tmp2); - tmp1 += tmp2; - data[i ] = tmp1 + csc; - data[n - i - 1] = tmp1 - csc; + tmp1 += tmp2; + data[i] = tmp1 + csc; + data[n - i - 1] = tmp1 - csc; } } @@ -136,40 +139,39 @@ static void ff_dct_calc_II_c(DCTContext *ctx, FFTSample *data) int i; float next; - for (i=0; i < n/2; i++) { - float tmp1 = data[i ]; + for (i = 0; i < n / 2; i++) { + float tmp1 = data[i]; float tmp2 = data[n - i - 1]; - float s = SIN(ctx, n, 2*i + 1); + float s = SIN(ctx, n, 2 * i + 1); - s *= tmp1 - tmp2; - tmp1 = (tmp1 + tmp2) * 0.5f; + s *= tmp1 - tmp2; + tmp1 = (tmp1 + tmp2) * 0.5f; - data[i ] = tmp1 + s; + data[i] = tmp1 + s; data[n-i-1] = tmp1 - s; } - ff_rdft_calc(&ctx->rdft, data); + ctx->rdft.rdft_calc(&ctx->rdft, data); - next = data[1] * 0.5; + next = data[1] * 0.5; data[1] *= -1; for (i = n - 2; i >= 0; i -= 2) { float inr = data[i ]; float ini = data[i + 1]; - float c = COS(ctx, n, i); - float s = SIN(ctx, n, i); - - data[i ] = c * inr + s * ini; + float c = COS(ctx, n, i); + float s = SIN(ctx, n, i); - data[i+1] = next; + data[i] = c * inr + s * ini; + data[i + 1] = next; - next += s * inr - c * ini; + next += s * inr - c * ini; } } -void ff_dct_calc(DCTContext *s, FFTSample *data) +static void dct32_func(DCTContext *ctx, FFTSample *data) { - s->dct_calc(s, data); + ctx->dct32(data, data); } av_cold int ff_dct_init(DCTContext *s, int nbits, enum DCTTransformType inverse) @@ -177,29 +179,39 @@ av_cold int ff_dct_init(DCTContext *s, int nbits, enum DCTTransformType inverse) int n = 1 << nbits; int i; - s->nbits = nbits; - s->inverse = inverse; + memset(s, 0, sizeof(*s)); - ff_init_ff_cos_tabs(nbits+2); + s->nbits = nbits; + s->inverse = inverse; - s->costab = ff_cos_tabs[nbits+2]; + if (inverse == DCT_II && nbits == 5) { + s->dct_calc = dct32_func; + } else { + ff_init_ff_cos_tabs(nbits + 2); - s->csc2 = av_malloc(n/2 * sizeof(FFTSample)); + s->costab = ff_cos_tabs[nbits + 2]; + s->csc2 = av_malloc(n / 2 * sizeof(FFTSample)); - if (ff_rdft_init(&s->rdft, nbits, inverse == DCT_III) < 0) { - av_free(s->csc2); - return -1; - } + if (ff_rdft_init(&s->rdft, nbits, inverse == DCT_III) < 0) { + av_free(s->csc2); + return -1; + } - for (i = 0; i < n/2; i++) - s->csc2[i] = 0.5 / sin((M_PI / (2*n) * (2*i + 1))); + for (i = 0; i < n / 2; i++) + s->csc2[i] = 0.5 / sin((M_PI / (2 * n) * (2 * i + 1))); - switch(inverse) { - case DCT_I : s->dct_calc = ff_dct_calc_I_c; break; - case DCT_II : s->dct_calc = ff_dct_calc_II_c ; break; - case DCT_III: s->dct_calc = ff_dct_calc_III_c; break; - case DST_I : s->dct_calc = ff_dst_calc_I_c; break; + switch (inverse) { + case DCT_I : s->dct_calc = ff_dct_calc_I_c; break; + case DCT_II : s->dct_calc = ff_dct_calc_II_c; break; + case DCT_III: s->dct_calc = ff_dct_calc_III_c; break; + case DST_I : s->dct_calc = ff_dst_calc_I_c; break; + } } + + s->dct32 = ff_dct32_float; + if (ARCH_X86) + ff_dct_init_x86(s); + return 0; }