MCKL
Monte Carlo Kernel Library
include
mckl
random
internal
philox_avx512_32_common.hpp
Go to the documentation of this file.
1
//============================================================================
2
// MCKL/include/mckl/random/internal/philox_avx512_32_common.hpp
3
//----------------------------------------------------------------------------
4
// MCKL: Monte Carlo Kernel Library
5
//----------------------------------------------------------------------------
6
// Copyright (c) 2013-2018, Yan Zhou
7
// All rights reserved.
8
//
9
// Redistribution and use in source and binary forms, with or without
10
// modification, are permitted provided that the following conditions are met:
11
//
12
// Redistributions of source code must retain the above copyright notice,
13
// this list of conditions and the following disclaimer.
14
//
15
// Redistributions in binary form must reproduce the above copyright notice,
16
// this list of conditions and the following disclaimer in the documentation
17
// and/or other materials provided with the distribution.
18
//
19
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS AS IS
20
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
23
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29
// POSSIBILITY OF SUCH DAMAGE.
30
//============================================================================
31
32
#ifndef MCKL_RANDOM_INTERNAL_PHILOX_AVX512_32_COMMON_HPP
33
#define MCKL_RANDOM_INTERNAL_PHILOX_AVX512_32_COMMON_HPP
34
35
#define MCKL_RANDOM_INTERNAL_PHILOX_AVX512_32_RBOX(K, N, imm8) \
36
{ \
37
constexpr int m0 = \
38
static_cast<int>(Constants::multiplier::value[0 % (K / 2)]); \
39
constexpr int m1 = \
40
static_cast<int>(Constants::multiplier::value[1 % (K / 2)]); \
41
constexpr int m2 = \
42
static_cast<int>(Constants::multiplier::value[2 % (K / 2)]); \
43
constexpr int m3 = \
44
static_cast<int>(Constants::multiplier::value[3 % (K / 2)]); \
45
constexpr int m4 = \
46
static_cast<int>(Constants::multiplier::value[4 % (K / 2)]); \
47
constexpr int m5 = \
48
static_cast<int>(Constants::multiplier::value[5 % (K / 2)]); \
49
constexpr int m6 = \
50
static_cast<int>(Constants::multiplier::value[6 % (K / 2)]); \
51
constexpr int m7 = \
52
static_cast<int>(Constants::multiplier::value[7 % (K / 2)]); \
53
const __m512i ymmm = _mm512_set_epi32( \
54
0, m7, 0, m6, 0, m5, 0, m4, 0, m3, 0, m2, 0, m1, 0, m0); \
55
const __m512i ymm8 = _mm512_mul_epu32(ymm0, ymmm); \
56
const __m512i ymm9 = _mm512_mul_epu32(ymm1, ymmm); \
57
const __m512i ymmA = _mm512_mul_epu32(ymm2, ymmm); \
58
const __m512i ymmB = _mm512_mul_epu32(ymm3, ymmm); \
59
const __m512i ymmC = _mm512_mul_epu32(ymm4, ymmm); \
60
const __m512i ymmD = _mm512_mul_epu32(ymm5, ymmm); \
61
const __m512i ymmE = _mm512_mul_epu32(ymm6, ymmm); \
62
const __m512i ymmF = _mm512_mul_epu32(ymm7, ymmm); \
63
\
64
constexpr int ma = static_cast<int>(0xFFFFFFFF); \
65
const __m512i ymma = _mm512_set_epi32( \
66
ma, 0, ma, 0, ma, 0, ma, 0, ma, 0, ma, 0, ma, 0, ma, 0); \
67
ymm0 = _mm512_and_si512(ymm0, ymma); \
68
ymm1 = _mm512_and_si512(ymm1, ymma); \
69
ymm2 = _mm512_and_si512(ymm2, ymma); \
70
ymm3 = _mm512_and_si512(ymm3, ymma); \
71
ymm4 = _mm512_and_si512(ymm4, ymma); \
72
ymm5 = _mm512_and_si512(ymm5, ymma); \
73
ymm6 = _mm512_and_si512(ymm6, ymma); \
74
ymm7 = _mm512_and_si512(ymm7, ymma); \
75
\
76
constexpr int w0 = \
77
static_cast<int>(Constants::weyl::value[0 % (K / 2)] * N); \
78
constexpr int w1 = \
79
static_cast<int>(Constants::weyl::value[1 % (K / 2)] * N); \
80
constexpr int w2 = \
81
static_cast<int>(Constants::weyl::value[2 % (K / 2)] * N); \
82
constexpr int w3 = \
83
static_cast<int>(Constants::weyl::value[3 % (K / 2)] * N); \
84
constexpr int w4 = \
85
static_cast<int>(Constants::weyl::value[4 % (K / 2)] * N); \
86
constexpr int w5 = \
87
static_cast<int>(Constants::weyl::value[5 % (K / 2)] * N); \
88
constexpr int w6 = \
89
static_cast<int>(Constants::weyl::value[6 % (K / 2)] * N); \
90
constexpr int w7 = \
91
static_cast<int>(Constants::weyl::value[7 % (K / 2)] * N); \
92
const __m512i ymmw = _mm512_set_epi32( \
93
w7, 0, w6, 0, w5, 0, w4, 0, w3, 0, w2, 0, w1, 0, w0, 0); \
94
const __m512i ymmk = _mm512_add_epi32(ymmk0, ymmw); \
95
ymm0 = _mm512_xor_si512(ymm0, ymmk); \
96
ymm1 = _mm512_xor_si512(ymm1, ymmk); \
97
ymm2 = _mm512_xor_si512(ymm2, ymmk); \
98
ymm3 = _mm512_xor_si512(ymm3, ymmk); \
99
ymm4 = _mm512_xor_si512(ymm4, ymmk); \
100
ymm5 = _mm512_xor_si512(ymm5, ymmk); \
101
ymm6 = _mm512_xor_si512(ymm6, ymmk); \
102
ymm7 = _mm512_xor_si512(ymm7, ymmk); \
103
\
104
ymm0 = _mm512_xor_si512(ymm0, ymm8); \
105
ymm1 = _mm512_xor_si512(ymm1, ymm9); \
106
ymm2 = _mm512_xor_si512(ymm2, ymmA); \
107
ymm3 = _mm512_xor_si512(ymm3, ymmB); \
108
ymm4 = _mm512_xor_si512(ymm4, ymmC); \
109
ymm5 = _mm512_xor_si512(ymm5, ymmD); \
110
ymm6 = _mm512_xor_si512(ymm6, ymmE); \
111
ymm7 = _mm512_xor_si512(ymm7, ymmF); \
112
\
113
ymm0 = _mm512_shuffle_epi32(ymm0, imm8); \
114
ymm1 = _mm512_shuffle_epi32(ymm1, imm8); \
115
ymm2 = _mm512_shuffle_epi32(ymm2, imm8); \
116
ymm3 = _mm512_shuffle_epi32(ymm3, imm8); \
117
ymm4 = _mm512_shuffle_epi32(ymm4, imm8); \
118
ymm5 = _mm512_shuffle_epi32(ymm5, imm8); \
119
ymm6 = _mm512_shuffle_epi32(ymm6, imm8); \
120
ymm7 = _mm512_shuffle_epi32(ymm7, imm8); \
121
}
122
123
#endif // MCKL_RANDOM_INTERNAL_PHILOX_AVX512_32_COMMON_HPP
Generated by
1.8.13
1.8.13