111
|
1 `/* Implementation of the MATMUL intrinsic
|
145
|
2 Copyright (C) 2002-2020 Free Software Foundation, Inc.
|
111
|
3 Contributed by Thomas Koenig <tkoenig@gcc.gnu.org>.
|
|
4
|
|
5 This file is part of the GNU Fortran runtime library (libgfortran).
|
|
6
|
|
7 Libgfortran is free software; you can redistribute it and/or
|
|
8 modify it under the terms of the GNU General Public
|
|
9 License as published by the Free Software Foundation; either
|
|
10 version 3 of the License, or (at your option) any later version.
|
|
11
|
|
12 Libgfortran is distributed in the hope that it will be useful,
|
|
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
15 GNU General Public License for more details.
|
|
16
|
|
17 Under Section 7 of GPL version 3, you are granted additional
|
|
18 permissions described in the GCC Runtime Library Exception, version
|
|
19 3.1, as published by the Free Software Foundation.
|
|
20
|
|
21 You should have received a copy of the GNU General Public License and
|
|
22 a copy of the GCC Runtime Library Exception along with this program;
|
|
23 see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
|
|
24 <http://www.gnu.org/licenses/>. */
|
|
25
|
|
26 #include "libgfortran.h"
|
|
27 #include <string.h>
|
|
28 #include <assert.h>'
|
|
29
|
|
30 include(iparm.m4)dnl
|
|
31
|
|
32 /* These are the specific versions of matmul with -mprefer-avx128. */
|
|
33
|
|
34 `#if defined (HAVE_'rtype_name`)
|
|
35
|
|
36 /* Prototype for the BLAS ?gemm subroutine, a pointer to which can be
|
|
37 passed to us by the front-end, in which case we call it for large
|
|
38 matrices. */
|
|
39
|
|
40 typedef void (*blas_call)(const char *, const char *, const int *, const int *,
|
|
41 const int *, const 'rtype_name` *, const 'rtype_name` *,
|
|
42 const int *, const 'rtype_name` *, const int *,
|
|
43 const 'rtype_name` *, 'rtype_name` *, const int *,
|
|
44 int, int);
|
|
45
|
|
46 #if defined(HAVE_AVX) && defined(HAVE_FMA3) && defined(HAVE_AVX128)
|
|
47 'define(`matmul_name',`matmul_'rtype_code`_avx128_fma3')dnl
|
|
48 `void
|
|
49 'matmul_name` ('rtype` * const restrict retarray,
|
|
50 'rtype` * const restrict a, 'rtype` * const restrict b, int try_blas,
|
|
51 int blas_limit, blas_call gemm) __attribute__((__target__("avx,fma")));
|
|
52 internal_proto('matmul_name`);
|
|
53 'include(matmul_internal.m4)dnl
|
|
54 `#endif
|
|
55
|
|
56 #if defined(HAVE_AVX) && defined(HAVE_FMA4) && defined(HAVE_AVX128)
|
|
57 'define(`matmul_name',`matmul_'rtype_code`_avx128_fma4')dnl
|
|
58 `void
|
|
59 'matmul_name` ('rtype` * const restrict retarray,
|
|
60 'rtype` * const restrict a, 'rtype` * const restrict b, int try_blas,
|
|
61 int blas_limit, blas_call gemm) __attribute__((__target__("avx,fma4")));
|
|
62 internal_proto('matmul_name`);
|
|
63 'include(matmul_internal.m4)dnl
|
|
64 `#endif
|
|
65
|
|
66 #endif
|
|
67 '
|