src/gromacs/simd/impl_ibm_vsx/impl_ibm_vsx_simd4_float.h

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2014,2015, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source 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
  36 #ifndef GMX_SIMD_IMPLEMENTATION_IBM_VSX_SIMD4_FLOAT_H
  37 #define GMX_SIMD_IMPLEMENTATION_IBM_VSX_SIMD4_FLOAT_H
  38
  39 #include <math.h>
  40
  41 #include <altivec.h>
  42
  43 #include "impl_ibm_vsx_common.h"
  44 #include "impl_ibm_vsx_simd_float.h"
  45
  46 /* IBM VSX SIMD instruction wrappers. Power7 and later.
  47  *
  48  * While this instruction set is similar to VMX, there are quite a few differences
  49  * that make it easier to understand if we start from scratch rather than by
  50  * including the VMX implementation and changing lots of things.
  51  */
  52
  53
  54 /* Make sure we do not screw up c++ - undefine vector/bool, and rely on __vector,
  55  * which is present both on gcc and xlc.
  56  */
  57 #undef vector
  58
  59 /* g++ is also unhappy with the clash of vector bool and the C++ reserved 'bool',
  60  * which is solved by undefining bool and reyling on __bool. However, that does
  61  * not work with xlc, which requires us to use bool. Solve the conflict by
  62  * defining a new vsx_bool.
  63  */
  64 #if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
  65 #    define vsx_bool __bool
  66 #    undef  bool
  67 #else
  68 #    define vsx_bool bool
  69 #endif
  70
  71
  72 /* Single precision VSX is 4 elements wide, use for SIMD4 */
  73 #define gmx_simd4_float_t                gmx_simd_float_t
  74 #define gmx_simd4_load_f                 gmx_simd_load_f
  75 #define gmx_simd4_load1_f                gmx_simd_load1_f
  76 #define gmx_simd4_set1_f                 gmx_simd_set1_f
  77 #define gmx_simd4_store_f                gmx_simd_store_f
  78 #define gmx_simd4_loadu_f                gmx_simd_loadu_f
  79 #define gmx_simd4_storeu_f               gmx_simd_storeu_f
  80 #define gmx_simd4_setzero_f              gmx_simd_setzero_f
  81 #define gmx_simd4_add_f                  gmx_simd_add_f
  82 #define gmx_simd4_sub_f                  gmx_simd_sub_f
  83 #define gmx_simd4_mul_f                  gmx_simd_mul_f
  84 #define gmx_simd4_fmadd_f                gmx_simd_fmadd_f
  85 #define gmx_simd4_fmsub_f                gmx_simd_fmsub_f
  86 #define gmx_simd4_fnmadd_f               gmx_simd_fnmadd_f
  87 #define gmx_simd4_fnmsub_f               gmx_simd_fnmsub_f
  88 #define gmx_simd4_and_f                  gmx_simd_and_f
  89 #define gmx_simd4_andnot_f               gmx_simd_andnot_f
  90 #define gmx_simd4_or_f                   gmx_simd_or_f
  91 #define gmx_simd4_xor_f                  gmx_simd_xor_f
  92 #define gmx_simd4_rsqrt_f                gmx_simd_rsqrt_f
  93 #define gmx_simd4_rcp_f                  gmx_simd_rcp_f
  94 #define gmx_simd4_fabs_f                 gmx_simd_fabs_f
  95 #define gmx_simd4_fneg_f                 gmx_simd_fneg_f
  96 #define gmx_simd4_max_f                  gmx_simd_max_f
  97 #define gmx_simd4_min_f                  gmx_simd_min_f
  98 #define gmx_simd4_round_f                gmx_simd_round_f
  99 #define gmx_simd4_trunc_f                gmx_simd_trunc_f
 100 #define gmx_simd4_fraction_f             gmx_simd_fraction_f
 101 #define gmx_simd4_get_exponent_f         gmx_simd_get_exponent_f
 102 #define gmx_simd4_get_mantissa_f         gmx_simd_get_mantissa_f
 103 #define gmx_simd4_set_exponent_f         gmx_simd_set_exponent_f
 104 #define gmx_simd4_dotproduct3_f          gmx_simd4_dotproduct3_f_ibm_vsx
 105 #define gmx_simd4_fint32_t               gmx_simd_fint32_t
 106 #define gmx_simd4_load_fi                gmx_simd_load_fi
 107 #define gmx_simd4_load1_fi               gmx_simd_load1_fi
 108 #define gmx_simd4_set1_fi                gmx_simd_set1_fi
 109 #define gmx_simd4_store_fi               gmx_simd_store_fi
 110 #define gmx_simd4_loadu_fi               gmx_simd_loadu_fi
 111 #define gmx_simd4_storeu_fi              gmx_simd_storeu_fi
 112 #define gmx_simd4_setzero_fi             gmx_simd_setzero_fi
 113 #define gmx_simd4_cvt_f2i                gmx_simd_cvt_f2i
 114 #define gmx_simd4_cvtt_f2i               gmx_simd_cvtt_f2i
 115 #define gmx_simd4_cvt_i2f                gmx_simd_cvt_i2f
 116 #define gmx_simd4_fbool_t                gmx_simd_fbool_t
 117 #define gmx_simd4_cmpeq_f                gmx_simd_cmpeq_f
 118 #define gmx_simd4_cmplt_f                gmx_simd_cmplt_f
 119 #define gmx_simd4_cmple_f                gmx_simd_cmple_f
 120 #define gmx_simd4_and_fb                 gmx_simd_and_fb
 121 #define gmx_simd4_or_fb                  gmx_simd_or_fb
 122 #define gmx_simd4_anytrue_fb             gmx_simd_anytrue_fb
 123 #define gmx_simd4_blendzero_f            gmx_simd_blendzero_f
 124 #define gmx_simd4_blendnotzero_f         gmx_simd_blendnotzero_f
 125 #define gmx_simd4_blendv_f               gmx_simd_blendv_f
 126 #define gmx_simd4_reduce_f               gmx_simd_reduce_f
 127
 128 static gmx_inline float
 129 gmx_simd4_dotproduct3_f_ibm_vsx(gmx_simd4_float_t a, gmx_simd4_float_t b)
 130 {
 131     gmx_simd4_float_t            c     = gmx_simd_mul_f(a, b);
 132     const __vector unsigned char perm1 = { 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7 };
 133     const __vector unsigned char perm2 = { 4, 5, 6, 7, 0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 2, 3 };
 134     __vector float               sum;
 135     sum = vec_add(c, vec_perm(c, c, (__vector unsigned char)perm1));
 136     sum = vec_add(sum, vec_perm(c, c, (__vector unsigned char)perm2));
 137     return vec_extract(sum, 0);
 138 }
 139
 140 #endif /* GMX_SIMD_IMPLEMENTATION_IBM_VSX_SIMD4_FLOAT_H */