1 /* Copyright (C) 2003 Free Software Foundation, Inc.
2
3 This file is part of GNU CC.
4
5 GNU CC is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation; either version 2, or (at your option)
8 any later version.
9
10 GNU CC is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
14
15 You should have received a copy of the GNU General Public License
16 along with GNU CC; see the file COPYING. If not, write to
17 the Free Software Foundation, 59 Temple Place - Suite 330,
18 Boston, MA 02111-1307, USA. */
19
20 /* As a special exception, if you include this header file into source
21 files compiled by GCC, this header file does not by itself cause
22 the resulting executable to be covered by the GNU General Public
23 License. This exception does not however invalidate any other
24 reasons why the executable file might be covered by the GNU General
25 Public License. */
26
27 /* Implemented from the specification included in the Intel C++ Compiler
28 User Guide and Reference, version 8.0. */
29
30 #ifndef _PMMINTRIN_H_INCLUDED
31 #define _PMMINTRIN_H_INCLUDED
32
33 #ifdef __SSE3__
34 #include <xmmintrin.h>
35 #include <emmintrin.h>
36
37 /* Additional bits in the MXCSR. */
38 #define _MM_DENORMALS_ZERO_MASK 0x0040
39 #define _MM_DENORMALS_ZERO_ON 0x0040
40 #define _MM_DENORMALS_ZERO_OFF 0x0000
41
42 #define _MM_SET_DENORMALS_ZERO_MODE(mode) \
43 _mm_setcsr ((_mm_getcsr () & ~_MM_DENORMALS_ZERO_MASK) | (mode))
44 #define _MM_GET_DENORMALS_ZERO_MODE() \
45 (_mm_getcsr() & _MM_DENORMALS_ZERO_MASK)
46
47 static __inline __m128
_mm_addsub_ps(__m128 __X,__m128 __Y)48 _mm_addsub_ps (__m128 __X, __m128 __Y)
49 {
50 return (__m128) __builtin_ia32_addsubps ((__v4sf)__X, (__v4sf)__Y);
51 }
52
53 static __inline __m128
_mm_hadd_ps(__m128 __X,__m128 __Y)54 _mm_hadd_ps (__m128 __X, __m128 __Y)
55 {
56 return (__m128) __builtin_ia32_haddps ((__v4sf)__X, (__v4sf)__Y);
57 }
58
59 static __inline __m128
_mm_hsub_ps(__m128 __X,__m128 __Y)60 _mm_hsub_ps (__m128 __X, __m128 __Y)
61 {
62 return (__m128) __builtin_ia32_hsubps ((__v4sf)__X, (__v4sf)__Y);
63 }
64
65 static __inline __m128
_mm_movehdup_ps(__m128 __X)66 _mm_movehdup_ps (__m128 __X)
67 {
68 return (__m128) __builtin_ia32_movshdup ((__v4sf)__X);
69 }
70
71 static __inline __m128
_mm_moveldup_ps(__m128 __X)72 _mm_moveldup_ps (__m128 __X)
73 {
74 return (__m128) __builtin_ia32_movsldup ((__v4sf)__X);
75 }
76
77 static __inline __m128d
_mm_addsub_pd(__m128d __X,__m128d __Y)78 _mm_addsub_pd (__m128d __X, __m128d __Y)
79 {
80 return (__m128d) __builtin_ia32_addsubpd ((__v2df)__X, (__v2df)__Y);
81 }
82
83 static __inline __m128d
_mm_hadd_pd(__m128d __X,__m128d __Y)84 _mm_hadd_pd (__m128d __X, __m128d __Y)
85 {
86 return (__m128d) __builtin_ia32_haddpd ((__v2df)__X, (__v2df)__Y);
87 }
88
89 static __inline __m128d
_mm_hsub_pd(__m128d __X,__m128d __Y)90 _mm_hsub_pd (__m128d __X, __m128d __Y)
91 {
92 return (__m128d) __builtin_ia32_hsubpd ((__v2df)__X, (__v2df)__Y);
93 }
94
95 static __inline __m128d
_mm_loaddup_pd(double const * __P)96 _mm_loaddup_pd (double const *__P)
97 {
98 return (__m128d) __builtin_ia32_loadddup (__P);
99 }
100
101 static __inline __m128d
_mm_movedup_pd(__m128d __X)102 _mm_movedup_pd (__m128d __X)
103 {
104 return (__m128d) __builtin_ia32_movddup ((__v2df)__X);
105 }
106
107 static __inline __m128i
_mm_lddqu_si128(__m128i const * __P)108 _mm_lddqu_si128 (__m128i const *__P)
109 {
110 return (__m128i) __builtin_ia32_lddqu ((char const *)__P);
111 }
112
113 #if 0
114 static __inline void
115 _mm_monitor (void const * __P, unsigned int __E, unsigned int __H)
116 {
117 __builtin_ia32_monitor (__P, __E, __H);
118 }
119
120 static __inline void
121 _mm_mwait (unsigned int __E, unsigned int __H)
122 {
123 __builtin_ia32_mwait (__E, __H);
124 }
125 #else
126 #define _mm_monitor(P, E, H) __builtin_ia32_monitor ((P), (E), (H))
127 #define _mm_mwait(E, H) __builtin_ia32_mwait ((E), (H))
128 #endif
129
130 #endif /* __SSE3__ */
131
132 #endif /* _PMMINTRIN_H_INCLUDED */
133