include/gmx_avx_double.h

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013, by the GROMACS development team, led by
   5  * David van der Spoel, Berk Hess, Erik Lindahl, and including many
   6  * others, as listed in the AUTHORS file in the top-level source
   7  * directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 #ifndef _gmx_avx_double_h_
  36 #define _gmx_avx_double_h_
  37
  38 /* We require AVX now! */
  39
  40 #include <immintrin.h> /* AVX */
  41
  42 static inline __m256d
  43 gmx_mm256_invsqrt_pd(__m256d x)
  44 {
  45     /* There is no double precision AVX rsqrt instruction.
  46      * But using a single precision rsqrt still gives the full precision.
  47      */
  48     const __m256d half    = _mm256_set_pd(0.5, 0.5, 0.5, 0.5);
  49     const __m256d three   = _mm256_set_pd(3.0, 3.0, 3.0, 3.0);
  50
  51     __m256d       lu = _mm256_cvtps_pd(_mm_rsqrt_ps(_mm256_cvtpd_ps(x)));
  52
  53     lu = _mm256_mul_pd(half, _mm256_mul_pd(_mm256_sub_pd(three, _mm256_mul_pd(_mm256_mul_pd(lu, lu), x)), lu));
  54     return _mm256_mul_pd(half, _mm256_mul_pd(_mm256_sub_pd(three, _mm256_mul_pd(_mm256_mul_pd(lu, lu), x)), lu));
  55 }
  56
  57 static inline __m256d
  58 gmx_mm256_calc_rsq_pd(__m256d dx, __m256d dy, __m256d dz)
  59 {
  60     return _mm256_add_pd( _mm256_add_pd( _mm256_mul_pd(dx, dx), _mm256_mul_pd(dy, dy) ), _mm256_mul_pd(dz, dz) );
  61 }
  62
  63 /* Normal sum of four xmm registers */
  64 #define gmx_mm256_sum4_pd(t0, t1, t2, t3)  _mm256_add_pd(_mm256_add_pd(t0, t1), _mm256_add_pd(t2, t3))
  65
  66 #endif /* gmx_avx_double_h_ */