2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2014,2015,2017,2018,2019,2020, 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.
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.
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.
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.
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.
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.
36 #ifndef GMX_SIMD_IMPLEMENTATION_IBM_VSX_DEFINITIONS_H
37 #define GMX_SIMD_IMPLEMENTATION_IBM_VSX_DEFINITIONS_H
39 // IBM VSX SIMD instruction wrappers. Power7 and later.
41 // While this instruction set is similar to VMX, there are quite a few differences
42 // that make it easier to understand if we start from scratch rather than by
43 // including the VMX implementation and changing lots of things.
47 #if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
48 // According to G++ documentation, when using VSX in C++ we
49 // must undefine vector & bool macros after including altivec.h
52 # define vsxBool __bool
54 // We cannot undefine bool on xlc, but somehow it works anyway
59 #define GMX_SIMD_HAVE_FLOAT 1
60 // GMX_SIMD_HAVE_DOUBLE is conditionally defined further down
61 #define GMX_SIMD_HAVE_LOADU 1
62 #define GMX_SIMD_HAVE_STOREU 1
63 #define GMX_SIMD_HAVE_LOGICAL 1
64 #define GMX_SIMD_HAVE_FMA 1
65 #define GMX_SIMD_HAVE_FINT32_EXTRACT 1
66 #define GMX_SIMD_HAVE_FINT32_LOGICAL 1
67 #define GMX_SIMD_HAVE_FINT32_ARITHMETICS 1
68 // GMX_SIMD_HAVE_DINT32_EXTRACT is conditionally defined further down
69 // GMX_SIMD_HAVE_DINT32_LOGICAL is conditionally defined further down
70 // GMX_SIMD_HAVE_DINT32_ARITHMETICS is conditionally defined further down
71 #define GMX_SIMD_HAVE_NATIVE_COPYSIGN_FLOAT 1
72 #define GMX_SIMD_HAVE_NATIVE_RSQRT_ITER_FLOAT 0
73 #define GMX_SIMD_HAVE_NATIVE_RCP_ITER_FLOAT 0
74 #define GMX_SIMD_HAVE_NATIVE_LOG_FLOAT 0
75 #define GMX_SIMD_HAVE_NATIVE_EXP2_FLOAT 0
76 #define GMX_SIMD_HAVE_NATIVE_EXP_FLOAT 0
77 #define GMX_SIMD_HAVE_NATIVE_COPYSIGN_DOUBLE 1
78 #define GMX_SIMD_HAVE_NATIVE_RSQRT_ITER_DOUBLE 0
79 #define GMX_SIMD_HAVE_NATIVE_RCP_ITER_DOUBLE 0
80 #define GMX_SIMD_HAVE_NATIVE_LOG_DOUBLE 0
81 #define GMX_SIMD_HAVE_NATIVE_EXP2_DOUBLE 0
82 #define GMX_SIMD_HAVE_NATIVE_EXP_DOUBLE 0
83 #define GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_FLOAT 1
84 // GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_DOUBLE is conditionally defined further down
85 #define GMX_SIMD_HAVE_HSIMD_UTIL_FLOAT 0 // No need for half-simd, width is 4
86 #define GMX_SIMD_HAVE_HSIMD_UTIL_DOUBLE 0 // No need for half-simd, width is 2
87 #define GMX_SIMD4_HAVE_FLOAT 1
88 #define GMX_SIMD4_HAVE_DOUBLE 0
90 // With GCC, only version 4.9 or later supports all parts of double precision VSX.
91 // We check explicitly for xlc, since that compiler appears to like pretending it is gcc,
92 // but there double precision seems to work fine.
93 #if defined(__ibmxl__) || defined(__xlC__) \
94 || !(defined(__GNUC__) && ((__GNUC__ < 4) || ((__GNUC__ == 4) && (__GNUC_MINOR__ < 9))))
95 # define GMX_SIMD_HAVE_DOUBLE 1
96 # define GMX_SIMD_HAVE_DINT32_EXTRACT 1
97 # define GMX_SIMD_HAVE_DINT32_LOGICAL 1
98 # define GMX_SIMD_HAVE_DINT32_ARITHMETICS 1
99 # define GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_DOUBLE 1
101 # define GMX_SIMD_HAVE_DOUBLE 0
102 # define GMX_SIMD_HAVE_DINT32_EXTRACT 0
103 # define GMX_SIMD_HAVE_DINT32_LOGICAL 0
104 # define GMX_SIMD_HAVE_DINT32_ARITHMETICS 0
105 # define GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_DOUBLE 0
108 // Implementation details
109 #define GMX_SIMD_FLOAT_WIDTH 4
110 #define GMX_SIMD_DOUBLE_WIDTH 2
111 #define GMX_SIMD_FINT32_WIDTH 4
112 #define GMX_SIMD_DINT32_WIDTH 2
113 #define GMX_SIMD4_WIDTH 4
114 #define GMX_SIMD_ALIGNMENT 16 // Bytes (4*single or 2*sdouble)
115 #define GMX_SIMD_RSQRT_BITS 14
116 #define GMX_SIMD_RCP_BITS 14
119 #endif /* GMX_SIMD_IMPLEMENTATION_IBM_VSX_COMMON_H */