yading@10: /*
yading@10:  * Copyright (c) 2012
yading@10:  *      MIPS Technologies, Inc., California.
yading@10:  *
yading@10:  * Redistribution and use in source and binary forms, with or without
yading@10:  * modification, are permitted provided that the following conditions
yading@10:  * are met:
yading@10:  * 1. Redistributions of source code must retain the above copyright
yading@10:  *    notice, this list of conditions and the following disclaimer.
yading@10:  * 2. Redistributions in binary form must reproduce the above copyright
yading@10:  *    notice, this list of conditions and the following disclaimer in the
yading@10:  *    documentation and/or other materials provided with the distribution.
yading@10:  * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
yading@10:  *    contributors may be used to endorse or promote products derived from
yading@10:  *    this software without specific prior written permission.
yading@10:  *
yading@10:  * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
yading@10:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
yading@10:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
yading@10:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
yading@10:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
yading@10:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
yading@10:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
yading@10:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
yading@10:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
yading@10:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
yading@10:  * SUCH DAMAGE.
yading@10:  *
yading@10:  * Author:  Nedeljko Babic (nbabic@mips.com)
yading@10:  *
yading@10:  * adaptive and fixed codebook vector operations for ACELP-based codecs
yading@10:  * optimized for MIPS
yading@10:  *
yading@10:  * This file is part of FFmpeg.
yading@10:  *
yading@10:  * FFmpeg 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:  * FFmpeg 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 FFmpeg; 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: /**
yading@10:  * @file
yading@10:  * Reference: libavcodec/acelp_vectors.c
yading@10:  */
yading@10: #include "config.h"
yading@10: #include "libavcodec/acelp_vectors.h"
yading@10: 
yading@10: #if HAVE_INLINE_ASM
yading@10: static void ff_weighted_vector_sumf_mips(
yading@10:                   float *out, const float *in_a, const float *in_b,
yading@10:                   float weight_coeff_a, float weight_coeff_b, int length)
yading@10: {
yading@10:     const float *a_end = in_a + length;
yading@10: 
yading@10:     /* loop unrolled two times */
yading@10:     __asm__ volatile (
yading@10:         "blez   %[length], ff_weighted_vector_sumf_end%=                     \n\t"
yading@10: 
yading@10:         "ff_weighted_vector_sumf_madd%=:                                     \n\t"
yading@10:         "lwc1   $f0,       0(%[in_a])                                        \n\t"
yading@10:         "lwc1   $f3,       4(%[in_a])                                        \n\t"
yading@10:         "lwc1   $f1,       0(%[in_b])                                        \n\t"
yading@10:         "lwc1   $f4,       4(%[in_b])                                        \n\t"
yading@10:         "mul.s  $f2,       %[weight_coeff_a], $f0                            \n\t"
yading@10:         "mul.s  $f5,       %[weight_coeff_a], $f3                            \n\t"
yading@10:         "madd.s $f2,       $f2,               %[weight_coeff_b], $f1         \n\t"
yading@10:         "madd.s $f5,       $f5,               %[weight_coeff_b], $f4         \n\t"
yading@10:         "addiu  %[in_a],   8                                                 \n\t"
yading@10:         "addiu  %[in_b],   8                                                 \n\t"
yading@10:         "swc1   $f2,       0(%[out])                                         \n\t"
yading@10:         "swc1   $f5,       4(%[out])                                         \n\t"
yading@10:         "addiu  %[out],    8                                                 \n\t"
yading@10:         "bne   %[in_a],    %[a_end],          ff_weighted_vector_sumf_madd%= \n\t"
yading@10: 
yading@10:         "ff_weighted_vector_sumf_end%=:                                      \n\t"
yading@10: 
yading@10:         : [out] "+r" (out), [in_a] "+r" (in_a),   [in_b] "+r" (in_b)
yading@10:         : [weight_coeff_a] "f" (weight_coeff_a),
yading@10:           [weight_coeff_b] "f" (weight_coeff_b),
yading@10:           [length] "r" (length), [a_end]"r"(a_end)
yading@10:         : "$f0", "$f1", "$f2", "$f3", "$f4", "$f5", "memory"
yading@10:     );
yading@10: }
yading@10: #endif /* HAVE_INLINE_ASM */
yading@10: 
yading@10: void ff_acelp_vectors_init_mips(ACELPVContext *c)
yading@10: {
yading@10: #if HAVE_INLINE_ASM
yading@10:     c->weighted_vector_sumf = ff_weighted_vector_sumf_mips;
yading@10: #endif
yading@10: }