root/libfftss/r4_sse3_n.c

DEFINITIONS

1. kern_r4_sse3_n_f
2. kern_r4_sse3_n_b
3. fftss_r4_sse3_n_f
4. fftss_r4_sse3_n_b

/*
 * Copyright (C) 2002-2007 Akira Nukada. All rights reserved.
 * Copyright (C) 2002-2007 The SSI Project. All rights reserved. 
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Neither the name of the project nor the names of its contributors 
 *       may be used to endorse or promote products derived from this software 
 *       without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE SSI PROJECT ``AS IS'' AND ANY EXPRESS 
 * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE SSI PROJECT BE LIABLE FOR ANY DIRECT,
 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
 *  STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
 * THE POSSIBILITY OF SUCH DAMAGE.
 *
 */
#include "libfftss.h"

#ifdef USE_SSE3
#ifdef HAVE_PMMINTRIN_H
#include <pmmintrin.h>
#endif
#ifdef HAVE_INTRIN_H
#include <intrin.h>
#endif

extern unsigned int fftss_sse2_cmz[];
extern unsigned int fftss_sse2_czm[];

static inline void
kern_r4_sse3_n_f(double *restrict o0,
           double *restrict o1,
           double *restrict o2,
           double *restrict o3,
           double *i0, 
           double *i1, 
           double *i2, 
           double *i3, 
           double *w,
           long bsize, long blocks)
{
  long i, j;
  __m128d zm, mz;
  zm = _mm_loadu_pd((double *)fftss_sse2_czm);
  mz = _mm_loadu_pd((double *)fftss_sse2_cmz);

  for (i = 0; i < blocks; i++) {
    __m128d w1, w2, w3;
    __m128d w1p, w2p, w3p;
    
    w1p = w1 = _mm_load_pd(w + 2 * i * bsize);
    /* w1 = _mm_xor_pd(w1, _mm_set_pd(-0.0, 0.0)); */
    w1 = _mm_xor_pd(w1, zm);
    w1p = _mm_shuffle_pd(w1p, w1p, 1);

    w2p = w2 = _mm_load_pd(w + 4 * i * bsize);
    /* w2 = _mm_xor_pd(w2, _mm_set_pd(-0.0, 0.0)); */
    w2 = _mm_xor_pd(w2, zm);
    w2p = _mm_shuffle_pd(w2p, w2p, 1);

    w3p = w3 = _mm_load_pd(w + 6 * i * bsize);
    /* w3 = _mm_xor_pd(w3, _mm_set_pd(-0.0, 0.0)); */
    w3 = _mm_xor_pd(w3, zm);
    w3p = _mm_shuffle_pd(w3p, w3p, 1);

    for (j = 0; j < bsize; j++) {
      __m128d p0, p1, p2, p3;
      __m128d d0, d1, d2, d3;
      
      d1 = _mm_loaddup_pd(i1 + 2 * j);
      d1 = _mm_mul_pd(d1, w1);
      p1 = _mm_loaddup_pd(i1 + 2 * j + 1);
      p1 = _mm_mul_pd(p1, w1p); 
      d1 = _mm_add_pd(d1, p1);

      d3 = _mm_loaddup_pd(i3 + 2 * j);
      d3 = _mm_mul_pd(d3, w3);
      p3 = _mm_loaddup_pd(i3 + 2 * j + 1);
      p3 = _mm_mul_pd(p3, w3p); 
      d3 = _mm_add_pd(d3, p3);

      d2 = _mm_loaddup_pd(i2 + 2 * j);
      d2 = _mm_mul_pd(d2, w2);
      p2 = _mm_loaddup_pd(i2 + 2 * j + 1);
      p2 = _mm_mul_pd(p2, w2p); 
      d2 = _mm_add_pd(d2, p2);

      d0 = _mm_load_pd(i0 + 2 * j);
      p0 = _mm_add_pd(d0, d2);
      p2 = _mm_sub_pd(d0, d2);
      p1 = _mm_add_pd(d1, d3);
      p3 = _mm_sub_pd(d1, d3);
      /* p3 = _mm_xor_pd(p3, _mm_set_sd(-0.0)); */
      p3 = _mm_xor_pd(p3, mz);
      p3 = _mm_shuffle_pd(p3, p3, 1);
      
      _mm_store_pd(o0 + 2 * j, _mm_add_pd(p0, p1));
      _mm_store_pd(o2 + 2 * j, _mm_sub_pd(p0, p1));
      _mm_store_pd(o1 + 2 * j, _mm_add_pd(p2, p3));
      _mm_store_pd(o3 + 2 * j, _mm_sub_pd(p2, p3));

    }
    i0 += 8 * bsize; i1 += 8 * bsize;
    i2 += 8 * bsize; i3 += 8 * bsize;
    o0 += 2 * bsize; o1 += 2 * bsize;
    o2 += 2 * bsize; o3 += 2 * bsize;
  }
}

static inline void 
kern_r4_sse3_n_b(double *restrict o0,
                   double *restrict o1,
                   double *restrict o2,
                   double *restrict o3,
                   double *i0, 
                   double *i1, 
                   double *i2, 
                   double *i3, 
                   double *w,
                   long bsize, long blocks)
{
  long i, j;
  __m128d zm, mz;
  zm = _mm_loadu_pd((double *)fftss_sse2_czm);
  mz = _mm_loadu_pd((double *)fftss_sse2_cmz);

  for (i = 0; i < blocks; i++) {
    __m128d w1, w2, w3;
    __m128d w1p, w2p, w3p;
    
    w1p = w1 = _mm_load_pd(w + 2 * i * bsize);
    w1p = _mm_shuffle_pd(w1p, w1p, 1);

    w2p = w2 = _mm_load_pd(w + 4 * i * bsize);
    w2p = _mm_shuffle_pd(w2p, w2p, 1);

    w3p = w3 = _mm_load_pd(w + 6 * i * bsize);
    w3p = _mm_shuffle_pd(w3p, w3p, 1);
    
    for (j = 0; j < bsize; j++) {
      __m128d p0, p1, p2, p3;
      __m128d d0, d1, d2, d3;

      d1 = _mm_loaddup_pd(i1 + 2 * j);
      d1 = _mm_mul_pd(d1, w1);
      p1 = _mm_loaddup_pd(i1 + 2 * j + 1);
      p1 = _mm_mul_pd(p1, w1p); 
      d1 = _mm_addsub_pd(d1, p1);

      d3 = _mm_loaddup_pd(i3 + 2 * j);      
      d3 = _mm_mul_pd(d3, w3);
      p3 = _mm_loaddup_pd(i3 + 2 * j + 1);
      p3 = _mm_mul_pd(p3, w3p); 
      d3 = _mm_addsub_pd(d3, p3);

      d2 = _mm_loaddup_pd(i2 + 2 * j);
      d2 = _mm_mul_pd(d2, w2);
      p2 = _mm_loaddup_pd(i2 + 2 * j + 1);
      p2 = _mm_mul_pd(p2, w2p); 
      d2 = _mm_addsub_pd(d2, p2);

      d0 = _mm_load_pd(i0 + 2 * j);
      p0 = _mm_add_pd(d0, d2);
      p2 = _mm_sub_pd(d0, d2);
      p1 = _mm_add_pd(d1, d3);
      p3 = _mm_sub_pd(d1, d3);
      /* p3 = _mm_xor_pd(p3, _mm_set_sd(-0.0)); */
      p3 = _mm_xor_pd(p3, mz);
      p3 = _mm_shuffle_pd(p3, p3, 1);
      
      _mm_store_pd(o0 + 2 * j, _mm_add_pd(p0, p1));
      _mm_store_pd(o2 + 2 * j, _mm_sub_pd(p0, p1));
      _mm_store_pd(o1 + 2 * j, _mm_sub_pd(p2, p3));
      _mm_store_pd(o3 + 2 * j, _mm_add_pd(p2, p3));


    }
    i0 += 8 * bsize; i1 += 8 * bsize;
    i2 += 8 * bsize; i3 += 8 * bsize;
    o0 += 2 * bsize; o1 += 2 * bsize;
    o2 += 2 * bsize; o3 += 2 * bsize;
  }
}

void fftss_r4_sse3_n_f(double *in, double *out, double *w,
                   long bsize, long blocks)
{
  kern_r4_sse3_n_f(out, out + bsize * blocks * 2, 
             out + bsize * blocks * 4, out + bsize * blocks * 6,
             in, in + bsize * 2,
             in + bsize * 4, in + bsize * 6,
             w, bsize, blocks);
}

void fftss_r4_sse3_n_b(double *in, double *out, double *w,
                   long bsize, long blocks)
{
  kern_r4_sse3_n_b(out, out + bsize * blocks * 2, 
             out + bsize * blocks * 4, out + bsize * blocks * 6,
             in, in + bsize * 2,
             in + bsize * 4, in + bsize * 6,
             w, bsize, blocks);
}
                       
#endif