FD.io VPP
v20.09-64-g4f7b92f0a
Vector Packet Processing
vector.h
Go to the documentation of this file.
1
/*
2
* Copyright (c) 2015 Cisco and/or its affiliates.
3
* Licensed under the Apache License, Version 2.0 (the "License");
4
* you may not use this file except in compliance with the License.
5
* You may obtain a copy of the License at:
6
*
7
* http://www.apache.org/licenses/LICENSE-2.0
8
*
9
* Unless required by applicable law or agreed to in writing, software
10
* distributed under the License is distributed on an "AS IS" BASIS,
11
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
* See the License for the specific language governing permissions and
13
* limitations under the License.
14
*/
15
/*
16
Copyright (c) 2005 Eliot Dresselhaus
17
18
Permission is hereby granted, free of charge, to any person obtaining
19
a copy of this software and associated documentation files (the
20
"Software"), to deal in the Software without restriction, including
21
without limitation the rights to use, copy, modify, merge, publish,
22
distribute, sublicense, and/or sell copies of the Software, and to
23
permit persons to whom the Software is furnished to do so, subject to
24
the following conditions:
25
26
The above copyright notice and this permission notice shall be
27
included in all copies or substantial portions of the Software.
28
29
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
30
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
31
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
32
NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
33
LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
34
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
35
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
36
*/
37
38
#ifndef included_clib_vector_h
39
#define included_clib_vector_h
40
41
#include <
vppinfra/clib.h
>
42
43
/* Vector types. */
44
45
#if defined (__MMX__) || defined (__IWMMXT__) || defined (__aarch64__) \
46
|| defined (__i686__)
47
#define CLIB_HAVE_VEC64
48
#endif
49
50
#if defined (__aarch64__) && defined(__ARM_NEON) || defined (__i686__)
51
#define CLIB_HAVE_VEC128
52
#endif
53
54
#if defined (__SSE4_2__) && __GNUC__ >= 4
55
#define CLIB_HAVE_VEC128
56
#endif
57
58
#if defined (__ALTIVEC__)
59
#define CLIB_HAVE_VEC128
60
#endif
61
62
#if defined (__AVX2__)
63
#define CLIB_HAVE_VEC256
64
#if defined (__clang__) && __clang_major__ < 4
65
#undef CLIB_HAVE_VEC256
66
#endif
67
#endif
68
69
#if defined (__AVX512BITALG__)
70
#define CLIB_HAVE_VEC512
71
#endif
72
73
/* 128 implies 64 */
74
#ifdef CLIB_HAVE_VEC128
75
#define CLIB_HAVE_VEC64
76
#endif
77
78
#define _vector_size(n) __attribute__ ((vector_size (n)))
79
#define _vector_size_unaligned(n) __attribute__ ((vector_size (n), __aligned__ (1)))
80
81
#define foreach_vec64i _(i,8,8) _(i,16,4) _(i,32,2)
82
#define foreach_vec64u _(u,8,8) _(u,16,4) _(u,32,2)
83
#define foreach_vec64f _(f,32,2)
84
#define foreach_vec128i _(i,8,16) _(i,16,8) _(i,32,4) _(i,64,2)
85
#define foreach_vec128u _(u,8,16) _(u,16,8) _(u,32,4) _(u,64,2)
86
#define foreach_vec128f _(f,32,4) _(f,64,2)
87
#define foreach_vec256i _(i,8,32) _(i,16,16) _(i,32,8) _(i,64,4)
88
#define foreach_vec256u _(u,8,32) _(u,16,16) _(u,32,8) _(u,64,4)
89
#define foreach_vec256f _(f,32,8) _(f,64,4)
90
#define foreach_vec512i _(i,8,64) _(i,16,32) _(i,32,16) _(i,64,8)
91
#define foreach_vec512u _(u,8,64) _(u,16,32) _(u,32,16) _(u,64,8)
92
#define foreach_vec512f _(f,32,16) _(f,64,8)
93
94
#if defined (CLIB_HAVE_VEC512)
95
#define foreach_int_vec foreach_vec64i foreach_vec128i foreach_vec256i foreach_vec512i
96
#define foreach_uint_vec foreach_vec64u foreach_vec128u foreach_vec256u foreach_vec512u
97
#define foreach_float_vec foreach_vec64f foreach_vec128f foreach_vec256f foreach_vec512f
98
#elif defined (CLIB_HAVE_VEC256)
99
#define foreach_int_vec foreach_vec64i foreach_vec128i foreach_vec256i
100
#define foreach_uint_vec foreach_vec64u foreach_vec128u foreach_vec256u
101
#define foreach_float_vec foreach_vec64f foreach_vec128f foreach_vec256f
102
#else
103
#define foreach_int_vec foreach_vec64i foreach_vec128i
104
#define foreach_uint_vec foreach_vec64u foreach_vec128u
105
#define foreach_float_vec foreach_vec64f foreach_vec128f
106
#endif
107
108
#define foreach_vec foreach_int_vec foreach_uint_vec foreach_float_vec
109
110
/* *INDENT-OFF* */
111
112
/* Type Definitions */
113
#define _(t,s,c) \
114
typedef t##s t##s##x##c _vector_size (s/8*c); \
115
typedef t##s t##s##x##c##u _vector_size_unaligned (s/8*c); \
116
typedef union { \
117
t##s##x##c as_##t##s##x##c; \
118
t##s as_##t##s[c]; \
119
} t##s##x##c##_union_t;
120
121
foreach_vec64i
foreach_vec64u
foreach_vec64f
122
foreach_vec128i
foreach_vec128u
foreach_vec128f
123
foreach_vec256i
foreach_vec256u
foreach_vec256f
124
foreach_vec512i
foreach_vec512u
foreach_vec512f
125
#undef _
126
127
/* Vector word sized types. */
128
#ifndef CLIB_VECTOR_WORD_BITS
129
#ifdef CLIB_HAVE_VEC128
130
#define CLIB_VECTOR_WORD_BITS 128
131
#else
132
#define CLIB_VECTOR_WORD_BITS 64
133
#endif
134
#endif
/* CLIB_VECTOR_WORD_BITS */
135
136
/* Vector word sized types. */
137
#if CLIB_VECTOR_WORD_BITS == 128
138
typedef
i8
i8x _vector_size (16);
139
typedef
i16
i16x _vector_size (16);
140
typedef
i32
i32x _vector_size (16);
141
typedef
i64
i64x _vector_size (16);
142
typedef
u8
u8x _vector_size (16);
143
typedef
u16
u16x _vector_size (16);
144
typedef
u32
u32x _vector_size (16);
145
typedef
u64
u64x _vector_size (16);
146
#endif
147
#if CLIB_VECTOR_WORD_BITS == 64
148
typedef
i8
i8x _vector_size (8);
149
typedef
i16
i16x _vector_size (8);
150
typedef
i32
i32x _vector_size (8);
151
typedef
i64
i64x _vector_size (8);
152
typedef
u8
u8x _vector_size (8);
153
typedef
u16
u16x _vector_size (8);
154
typedef
u32
u32x _vector_size (8);
155
typedef
u64
u64x _vector_size (8);
156
#endif
157
158
/* universal inlines */
159
#define _(t, s, c) \
160
static_always_inline t##s##x##c \
161
t##s##x##c##_zero () \
162
{ return (t##s##x##c) {}; } \
163
164
foreach_vec
165
#undef _
166
167
#undef _vector_size
168
169
#define VECTOR_WORD_TYPE(t) t##x
170
#define VECTOR_WORD_TYPE_LEN(t) (sizeof (VECTOR_WORD_TYPE(t)) / sizeof (t))
171
172
#if defined (__SSE4_2__) && __GNUC__ >= 4
173
#include <
vppinfra/vector_sse42.h
>
174
#endif
175
176
#if defined (__AVX2__)
177
#include <
vppinfra/vector_avx2.h
>
178
#endif
179
180
#if defined (__AVX512BITALG__)
181
/* Due to power level transition issues, we don't preffer AVX-512 on
182
Skylake X and CascadeLake CPUs, AVX512BITALG is introduced on
183
icelake CPUs */
184
#include <
vppinfra/vector_avx512.h
>
185
#endif
186
187
#if defined (__ALTIVEC__)
188
#include <
vppinfra/vector_altivec.h
>
189
#endif
190
191
#if defined (__aarch64__)
192
#include <
vppinfra/vector_neon.h
>
193
#endif
194
195
#if (defined(CLIB_HAVE_VEC128) || defined(CLIB_HAVE_VEC64))
196
#include <
vppinfra/vector_funcs.h
>
197
#endif
198
199
/* this macro generate _splat inline functions for each scalar vector type */
200
#ifndef CLIB_VEC128_SPLAT_DEFINED
201
#define _(t, s, c) \
202
static_always_inline t##s##x##c \
203
t##s##x##c##_splat (t##s x) \
204
{ \
205
t##s##x##c r; \
206
int i; \
207
\
208
for (i = 0; i < c; i++) \
209
r[i] = x; \
210
\
211
return r; \
212
}
213
foreach_vec128i
foreach_vec128u
214
#undef _
215
#endif
216
217
/* *INDENT-ON* */
218
219
#endif
/* included_clib_vector_h */
220
/*
221
* fd.io coding-style-patch-verification: ON
222
*
223
* Local Variables:
224
* eval: (c-set-style "gnu")
225
* End:
226
*/
foreach_vec256u
#define foreach_vec256u
Definition:
vector.h:88
vector_sse42.h
u64
unsigned long u64
Definition:
types.h:89
foreach_vec128u
#define foreach_vec128u
Definition:
vector.h:85
u8
unsigned char u8
Definition:
types.h:56
foreach_vec512f
#define foreach_vec512f
Definition:
vector.h:92
foreach_vec128i
#define foreach_vec128i
Definition:
vector.h:84
foreach_vec128f
#define foreach_vec128f
Definition:
vector.h:86
vector_altivec.h
u32
unsigned int u32
Definition:
types.h:88
foreach_vec256f
#define foreach_vec256f
Definition:
vector.h:89
vector_avx2.h
u16
unsigned short u16
Definition:
types.h:57
i64
signed long i64
Definition:
types.h:78
i8
signed char i8
Definition:
types.h:45
clib.h
vector_avx512.h
vector_neon.h
foreach_vec512i
#define foreach_vec512i
Definition:
vector.h:90
i32
signed int i32
Definition:
types.h:77
foreach_vec64i
#define foreach_vec64i
Definition:
vector.h:81
foreach_vec512u
#define foreach_vec512u
Definition:
vector.h:91
foreach_vec256i
#define foreach_vec256i
Definition:
vector.h:87
vector_funcs.h
foreach_vec64f
#define foreach_vec64f
Definition:
vector.h:83
foreach_vec
#define foreach_vec
Definition:
vector.h:108
foreach_vec64u
#define foreach_vec64u
Definition:
vector.h:82
i16
signed short i16
Definition:
types.h:46
src
vppinfra
vector.h
Generated on Thu Jun 3 2021 16:07:06 for FD.io VPP by
1.8.13