/* -*- c++ -*- */

/*

 * Copyright 2002 Free Software Foundation, Inc.

 * 

 * This file is part of GNU Radio

 * 

 * GNU Radio is free software; you can redistribute it and/or modify

 * it under the terms of the GNU General Public License as published by

 * the Free Software Foundation; either version 3, or (at your option)

 * any later version.

 * 

 * GNU Radio 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 General Public License for more details.

 * 

 * You should have received a copy of the GNU General Public License

 * along with GNU Radio; see the file COPYING.  If not, write to

 * the Free Software Foundation, Inc., 51 Franklin Street,

 * Boston, MA 02110-1301, USA.

 */

#ifdef HAVE_CONFIG_H

#include <config.h>

#endif

#include <gr_fir_fff_simd.h>

#include <assert.h>

#include <malloc16.h>

#include <iostream>

using std::cerr;

gr_fir_fff_simd::gr_fir_fff_simd ()

  : gr_fir_fff_generic ()

{

  // cerr << "@@@ gr_fir_fff_simd\n";

  d_float_dotprod = 0;

  d_aligned_taps[0] = 0;

  d_aligned_taps[1] = 0;

  d_aligned_taps[2] = 0;

  d_aligned_taps[3] = 0;

}

gr_fir_fff_simd::gr_fir_fff_simd (const std::vector<float> &new_taps)

  : gr_fir_fff_generic (new_taps)

{

  // cerr << "@@@ gr_fir_fff_simd\n";

  d_float_dotprod = 0;

  d_aligned_taps[0] = 0;

  d_aligned_taps[1] = 0;

  d_aligned_taps[2] = 0;

  d_aligned_taps[3] = 0;

  set_taps (new_taps);

}

gr_fir_fff_simd::~gr_fir_fff_simd ()

{

  free16Align (d_aligned_taps[0]);

  free16Align (d_aligned_taps[1]);

  free16Align (d_aligned_taps[2]);

  free16Align (d_aligned_taps[3]);

}

void

gr_fir_fff_simd::set_taps (const std::vector<float> &inew_taps)

{

  gr_fir_fff::set_taps (inew_taps);        // call superclass

  const std::vector<float> new_taps = gr_reverse(inew_taps);

  unsigned len = new_taps.size ();

  // Make 4 copies of the coefficients, one for each data alignment

  // Note use of special 16-byte-aligned version of calloc()

  for (unsigned i = 0; i < 4; i++){

    free16Align (d_aligned_taps[i]);        // free old value

    // this works because the bit representation of a IEEE floating point

    // +zero is all zeros.  If you're using a different representation,

    // you'll need to explictly set the result to the appropriate 0.0 value.

    d_aligned_taps[i] = (float *) calloc16Align (1 + (len + i - 1) / 4,

                                               4 * sizeof (float));

    if (d_aligned_taps[i] == 0){

      // throw something...

      cerr << "@@@ gr_fir_fff_simd d_aligned_taps[" << i << "] == 0\n";

    }

    for (unsigned j = 0; j < len; j++)

      d_aligned_taps[i][j+i] = new_taps[j];

  }

}

float 

gr_fir_fff_simd::filter (const float input[])

{

  if (ntaps () == 0)

    return 0.0;

  // Round input data address down to 16 byte boundary

  // NB: depending on the alignment of input[], memory

  // before input[] will be accessed. The contents don't matter since 

  // they'll be multiplied by zero coefficients. I can't conceive of any

  // situation where this could cause a segfault since memory protection

  // in the x86 machines is done on much larger boundaries.

  const float *ar = (float *)((unsigned long) input & ~15);

  // Choose one of 4 sets of pre-shifted coefficients. al is both the

  // index into d_aligned_taps[] and the number of 0 words padded onto

  // that coefficients array for alignment purposes.

  unsigned al = input - ar;

  // call assembler routine to do the work, passing number of 4-float blocks.

  // assert (((unsigned long) ar & 15) == 0);

  // assert (((unsigned long) d_aligned_taps[al] & 15) == 0);

  // cerr << "ar: " << ar << " d_aligned_taps[ar]: " << d_aligned_taps[al]

  // << " (ntaps() + al - 1)/4 + 1: " << (ntaps() + al -1) / 4 + 1 << endl;

  float r = d_float_dotprod (ar, d_aligned_taps[al], (ntaps() + al - 1) / 4 + 1);

  // cerr << "result = " << r << endl;

  return r;

}