From fde82ca7e4e72226da99d86a9fa9689136461d14 Mon Sep 17 00:00:00 2001 From: Justin Ruggles Date: Sat, 17 Oct 2009 21:00:39 +0000 Subject: Move autocorrelation function from flacenc.c to lpc.c. Also rename the corresponding dsputil functions and remove their dependency on the FLAC encoder. Fixes Issue1486. Originally committed as revision 20266 to svn://svn.ffmpeg.org/ffmpeg/trunk --- libavcodec/Makefile | 2 +- libavcodec/dsputil.c | 8 +-- libavcodec/dsputil.h | 2 +- libavcodec/flacenc.c | 63 ------------------ libavcodec/lpc.c | 64 +++++++++++++++++- libavcodec/x86/dsputilenc_mmx.c | 6 +- libavcodec/x86/flacdsp_mmx.c | 139 ---------------------------------------- libavcodec/x86/lpc_mmx.c | 139 ++++++++++++++++++++++++++++++++++++++++ 8 files changed, 209 insertions(+), 214 deletions(-) delete mode 100644 libavcodec/x86/flacdsp_mmx.c create mode 100644 libavcodec/x86/lpc_mmx.c diff --git a/libavcodec/Makefile b/libavcodec/Makefile index 789ece0be0..85acb5f202 100644 --- a/libavcodec/Makefile +++ b/libavcodec/Makefile @@ -456,7 +456,6 @@ YASM-OBJS-$(CONFIG_GPL) += x86/h264_deblock_sse2.o \ MMX-OBJS-$(CONFIG_CAVS_DECODER) += x86/cavsdsp_mmx.o MMX-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc_mmx.o -MMX-OBJS-$(CONFIG_FLAC_ENCODER) += x86/flacdsp_mmx.o MMX-OBJS-$(CONFIG_GPL) += x86/idct_mmx.o MMX-OBJS-$(CONFIG_SNOW_DECODER) += x86/snowdsp_mmx.o MMX-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_mmx.o @@ -474,6 +473,7 @@ OBJS-$(HAVE_MMX) += x86/cpuid.o \ x86/fft.o \ x86/idct_mmx_xvid.o \ x86/idct_sse2_xvid.o \ + x86/lpc_mmx.o \ x86/motion_est_mmx.o \ x86/mpegvideo_mmx.o \ x86/simple_idct_mmx.o \ diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index be35ec3cff..aae86bcb21 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -45,8 +45,8 @@ void vorbis_inverse_coupling(float *mag, float *ang, int blocksize); /* ac3dec.c */ void ff_ac3_downmix_c(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len); -/* flacenc.c */ -void ff_flac_compute_autocorr(const int32_t *data, int len, int lag, double *autoc); +/* lpc.c */ +void ff_lpc_compute_autocorr(const int32_t *data, int len, int lag, double *autoc); /* pngdec.c */ void ff_add_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp); @@ -4837,9 +4837,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx) #if CONFIG_AC3_DECODER c->ac3_downmix = ff_ac3_downmix_c; #endif -#if CONFIG_FLAC_ENCODER - c->flac_compute_autocorr = ff_flac_compute_autocorr; -#endif + c->lpc_compute_autocorr = ff_lpc_compute_autocorr; c->vector_fmul = vector_fmul_c; c->vector_fmul_reverse = vector_fmul_reverse_c; c->vector_fmul_add = vector_fmul_add_c; diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h index 6cb5af2cc5..58524b26e4 100644 --- a/libavcodec/dsputil.h +++ b/libavcodec/dsputil.h @@ -386,7 +386,7 @@ typedef struct DSPContext { void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize); void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len); /* no alignment needed */ - void (*flac_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc); + void (*lpc_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc); /* assume len is a multiple of 8, and arrays are 16-byte aligned */ void (*vector_fmul)(float *dst, const float *src, int len); void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len); diff --git a/libavcodec/flacenc.c b/libavcodec/flacenc.c index 81a35a0d54..c1f862a684 100644 --- a/libavcodec/flacenc.c +++ b/libavcodec/flacenc.c @@ -552,69 +552,6 @@ static uint32_t calc_rice_params_lpc(RiceContext *rc, int pmin, int pmax, return bits; } -/** - * Apply Welch window function to audio block - */ -static void apply_welch_window(const int32_t *data, int len, double *w_data) -{ - int i, n2; - double w; - double c; - - assert(!(len&1)); //the optimization in r11881 does not support odd len - //if someone wants odd len extend the change in r11881 - - n2 = (len >> 1); - c = 2.0 / (len - 1.0); - - w_data+=n2; - data+=n2; - for(i=0; i 0); diff --git a/libavcodec/lpc.c b/libavcodec/lpc.c index 896db51759..49e41d8c34 100644 --- a/libavcodec/lpc.c +++ b/libavcodec/lpc.c @@ -26,6 +26,68 @@ #include "lpc.h" +/** + * Apply Welch window function to audio block + */ +static void apply_welch_window(const int32_t *data, int len, double *w_data) +{ + int i, n2; + double w; + double c; + + assert(!(len&1)); //the optimization in r11881 does not support odd len + //if someone wants odd len extend the change in r11881 + + n2 = (len >> 1); + c = 2.0 / (len - 1.0); + + w_data+=n2; + data+=n2; + for(i=0; i= MIN_LPC_ORDER && max_order <= MAX_LPC_ORDER && use_lpc > 0); if(use_lpc == 1){ - s->flac_compute_autocorr(samples, blocksize, max_order, autoc); + s->lpc_compute_autocorr(samples, blocksize, max_order, autoc); compute_lpc_coefs(autoc, max_order, &lpc[0][0], MAX_LPC_ORDER, 0, 1); diff --git a/libavcodec/x86/dsputilenc_mmx.c b/libavcodec/x86/dsputilenc_mmx.c index 6059436252..2bf0fe1328 100644 --- a/libavcodec/x86/dsputilenc_mmx.c +++ b/libavcodec/x86/dsputilenc_mmx.c @@ -1348,8 +1348,7 @@ static int ssd_int8_vs_int16_mmx(const int8_t *pix1, const int16_t *pix2, int si #endif //HAVE_SSSE3 -/* FLAC specific */ -void ff_flac_compute_autocorr_sse2(const int32_t *data, int len, int lag, +void ff_lpc_compute_autocorr_sse2(const int32_t *data, int len, int lag, double *autoc); @@ -1414,8 +1413,7 @@ void dsputilenc_init_mmx(DSPContext* c, AVCodecContext *avctx) c->sum_abs_dctelem= sum_abs_dctelem_sse2; c->hadamard8_diff[0]= hadamard8_diff16_sse2; c->hadamard8_diff[1]= hadamard8_diff_sse2; - if (CONFIG_FLAC_ENCODER) - c->flac_compute_autocorr = ff_flac_compute_autocorr_sse2; + c->lpc_compute_autocorr = ff_lpc_compute_autocorr_sse2; } #if HAVE_SSSE3 diff --git a/libavcodec/x86/flacdsp_mmx.c b/libavcodec/x86/flacdsp_mmx.c deleted file mode 100644 index 01c0d7ae8a..0000000000 --- a/libavcodec/x86/flacdsp_mmx.c +++ /dev/null @@ -1,139 +0,0 @@ -/* - * MMX optimized FLAC DSP utils - * Copyright (c) 2007 Loren Merritt - * - * This file is part of FFmpeg. - * - * FFmpeg 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, - * 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 - * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - */ - -#include "libavutil/x86_cpu.h" -#include "dsputil_mmx.h" - -static void apply_welch_window_sse2(const int32_t *data, int len, double *w_data) -{ - double c = 2.0 / (len-1.0); - int n2 = len>>1; - x86_reg i = -n2*sizeof(int32_t); - x86_reg j = n2*sizeof(int32_t); - __asm__ volatile( - "movsd %0, %%xmm7 \n\t" - "movapd "MANGLE(ff_pd_1)", %%xmm6 \n\t" - "movapd "MANGLE(ff_pd_2)", %%xmm5 \n\t" - "movlhps %%xmm7, %%xmm7 \n\t" - "subpd %%xmm5, %%xmm7 \n\t" - "addsd %%xmm6, %%xmm7 \n\t" - ::"m"(c) - ); -#define WELCH(MOVPD, offset)\ - __asm__ volatile(\ - "1: \n\t"\ - "movapd %%xmm7, %%xmm1 \n\t"\ - "mulpd %%xmm1, %%xmm1 \n\t"\ - "movapd %%xmm6, %%xmm0 \n\t"\ - "subpd %%xmm1, %%xmm0 \n\t"\ - "pshufd $0x4e, %%xmm0, %%xmm1 \n\t"\ - "cvtpi2pd (%3,%0), %%xmm2 \n\t"\ - "cvtpi2pd "#offset"*4(%3,%1), %%xmm3 \n\t"\ - "mulpd %%xmm0, %%xmm2 \n\t"\ - "mulpd %%xmm1, %%xmm3 \n\t"\ - "movapd %%xmm2, (%2,%0,2) \n\t"\ - MOVPD" %%xmm3, "#offset"*8(%2,%1,2) \n\t"\ - "subpd %%xmm5, %%xmm7 \n\t"\ - "sub $8, %1 \n\t"\ - "add $8, %0 \n\t"\ - "jl 1b \n\t"\ - :"+&r"(i), "+&r"(j)\ - :"r"(w_data+n2), "r"(data+n2)\ - ); - if(len&1) - WELCH("movupd", -1) - else - WELCH("movapd", -2) -#undef WELCH -} - -void ff_flac_compute_autocorr_sse2(const int32_t *data, int len, int lag, - double *autoc) -{ - double tmp[len + lag + 2]; - double *data1 = tmp + lag; - int j; - - if((x86_reg)data1 & 15) - data1++; - - apply_welch_window_sse2(data, len, data1); - - for(j=0; j>1; + x86_reg i = -n2*sizeof(int32_t); + x86_reg j = n2*sizeof(int32_t); + __asm__ volatile( + "movsd %0, %%xmm7 \n\t" + "movapd "MANGLE(ff_pd_1)", %%xmm6 \n\t" + "movapd "MANGLE(ff_pd_2)", %%xmm5 \n\t" + "movlhps %%xmm7, %%xmm7 \n\t" + "subpd %%xmm5, %%xmm7 \n\t" + "addsd %%xmm6, %%xmm7 \n\t" + ::"m"(c) + ); +#define WELCH(MOVPD, offset)\ + __asm__ volatile(\ + "1: \n\t"\ + "movapd %%xmm7, %%xmm1 \n\t"\ + "mulpd %%xmm1, %%xmm1 \n\t"\ + "movapd %%xmm6, %%xmm0 \n\t"\ + "subpd %%xmm1, %%xmm0 \n\t"\ + "pshufd $0x4e, %%xmm0, %%xmm1 \n\t"\ + "cvtpi2pd (%3,%0), %%xmm2 \n\t"\ + "cvtpi2pd "#offset"*4(%3,%1), %%xmm3 \n\t"\ + "mulpd %%xmm0, %%xmm2 \n\t"\ + "mulpd %%xmm1, %%xmm3 \n\t"\ + "movapd %%xmm2, (%2,%0,2) \n\t"\ + MOVPD" %%xmm3, "#offset"*8(%2,%1,2) \n\t"\ + "subpd %%xmm5, %%xmm7 \n\t"\ + "sub $8, %1 \n\t"\ + "add $8, %0 \n\t"\ + "jl 1b \n\t"\ + :"+&r"(i), "+&r"(j)\ + :"r"(w_data+n2), "r"(data+n2)\ + ); + if(len&1) + WELCH("movupd", -1) + else + WELCH("movapd", -2) +#undef WELCH +} + +void ff_lpc_compute_autocorr_sse2(const int32_t *data, int len, int lag, + double *autoc) +{ + double tmp[len + lag + 2]; + double *data1 = tmp + lag; + int j; + + if((x86_reg)data1 & 15) + data1++; + + apply_welch_window_sse2(data, len, data1); + + for(j=0; j