yading@10: /* yading@10: * AAC Spectral Band Replication decoding functions yading@10: * Copyright (c) 2012 Christophe Gisquet yading@10: * yading@10: * This file is part of Libav. yading@10: * yading@10: * Libav is free software; you can redistribute it and/or yading@10: * modify it under the terms of the GNU Lesser General Public yading@10: * License as published by the Free Software Foundation; either yading@10: * version 2.1 of the License, or (at your option) any later version. yading@10: * yading@10: * Libav is distributed in the hope that it will be useful, yading@10: * but WITHOUT ANY WARRANTY; without even the implied warranty of yading@10: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU yading@10: * Lesser General Public License for more details. yading@10: * yading@10: * You should have received a copy of the GNU Lesser General Public yading@10: * License along with Libav; if not, write to the Free Software yading@10: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA yading@10: */ yading@10: yading@10: #include "config.h" yading@10: #include "libavutil/attributes.h" yading@10: #include "libavutil/cpu.h" yading@10: #include "libavutil/x86/cpu.h" yading@10: #include "libavcodec/sbrdsp.h" yading@10: yading@10: float ff_sbr_sum_square_sse(float (*x)[2], int n); yading@10: void ff_sbr_sum64x5_sse(float *z); yading@10: void ff_sbr_hf_g_filt_sse(float (*Y)[2], const float (*X_high)[40][2], yading@10: const float *g_filt, int m_max, intptr_t ixh); yading@10: void ff_sbr_hf_gen_sse(float (*X_high)[2], const float (*X_low)[2], yading@10: const float alpha0[2], const float alpha1[2], yading@10: float bw, int start, int end); yading@10: void ff_sbr_neg_odd_64_sse(float *z); yading@10: void ff_sbr_qmf_post_shuffle_sse(float W[32][2], const float *z); yading@10: void ff_sbr_qmf_deint_bfly_sse(float *v, const float *src0, const float *src1); yading@10: void ff_sbr_qmf_deint_bfly_sse2(float *v, const float *src0, const float *src1); yading@10: void ff_sbr_qmf_pre_shuffle_sse2(float *z); yading@10: yading@10: void ff_sbr_hf_apply_noise_0_sse2(float (*Y)[2], const float *s_m, yading@10: const float *q_filt, int noise, yading@10: int kx, int m_max); yading@10: void ff_sbr_hf_apply_noise_1_sse2(float (*Y)[2], const float *s_m, yading@10: const float *q_filt, int noise, yading@10: int kx, int m_max); yading@10: void ff_sbr_hf_apply_noise_2_sse2(float (*Y)[2], const float *s_m, yading@10: const float *q_filt, int noise, yading@10: int kx, int m_max); yading@10: void ff_sbr_hf_apply_noise_3_sse2(float (*Y)[2], const float *s_m, yading@10: const float *q_filt, int noise, yading@10: int kx, int m_max); yading@10: yading@10: av_cold void ff_sbrdsp_init_x86(SBRDSPContext *s) yading@10: { yading@10: int mm_flags = av_get_cpu_flags(); yading@10: yading@10: if (EXTERNAL_SSE(mm_flags)) { yading@10: s->neg_odd_64 = ff_sbr_neg_odd_64_sse; yading@10: s->sum_square = ff_sbr_sum_square_sse; yading@10: s->sum64x5 = ff_sbr_sum64x5_sse; yading@10: s->hf_g_filt = ff_sbr_hf_g_filt_sse; yading@10: s->hf_gen = ff_sbr_hf_gen_sse; yading@10: s->qmf_post_shuffle = ff_sbr_qmf_post_shuffle_sse; yading@10: s->qmf_deint_bfly = ff_sbr_qmf_deint_bfly_sse; yading@10: } yading@10: yading@10: if (EXTERNAL_SSE2(mm_flags)) { yading@10: s->qmf_deint_bfly = ff_sbr_qmf_deint_bfly_sse2; yading@10: s->qmf_pre_shuffle = ff_sbr_qmf_pre_shuffle_sse2; yading@10: s->hf_apply_noise[0] = ff_sbr_hf_apply_noise_0_sse2; yading@10: s->hf_apply_noise[1] = ff_sbr_hf_apply_noise_1_sse2; yading@10: s->hf_apply_noise[2] = ff_sbr_hf_apply_noise_2_sse2; yading@10: s->hf_apply_noise[3] = ff_sbr_hf_apply_noise_3_sse2; yading@10: } yading@10: }