1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
/*******************************************************************************
* Copyright 2019 Intel Corporation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*******************************************************************************/
#ifndef COMMON_BFLOAT16_HPP
#define COMMON_BFLOAT16_HPP
#include <array>
#include <cmath>
#include <cstddef>
#include <cstdint>
#include <cstdlib>
#include <limits>
#include <type_traits>
#include "common/bit_cast.hpp"
#include "oneapi/dnnl/dnnl.h"
namespace dnnl {
namespace impl {
#if DNNL_CPU_RUNTIME != DNNL_RUNTIME_NONE
struct bfloat16_t;
bool try_cvt_float_to_bfloat16(bfloat16_t *out, const float *inp);
#endif
struct bfloat16_t {
uint16_t raw_bits_;
bfloat16_t() = default;
constexpr bfloat16_t(uint16_t r, bool) : raw_bits_(r) {}
bfloat16_t(float f) { (*this) = f; }
template <typename IntegerType,
typename SFINAE = typename std::enable_if<
std::is_integral<IntegerType>::value>::type>
bfloat16_t(const IntegerType i)
: raw_bits_ {convert_bits_of_normal_or_zero(
utils::bit_cast<uint32_t>(static_cast<float>(i)))} {}
bfloat16_t DNNL_API &operator=(float f);
template <typename IntegerType,
typename SFINAE = typename std::enable_if<
std::is_integral<IntegerType>::value>::type>
bfloat16_t &operator=(const IntegerType i) {
// Call the converting constructor that is optimized for integer types,
// followed by the fast defaulted move-assignment operator.
return (*this) = bfloat16_t {i};
}
DNNL_API operator float() const;
bfloat16_t &operator+=(const float a) {
(*this) = float {*this} + a;
return *this;
}
private:
// Converts the 32 bits of a normal float or zero to the bits of a bfloat16.
static constexpr uint16_t convert_bits_of_normal_or_zero(
const uint32_t bits) {
return static_cast<uint16_t>(
uint32_t {bits
+ uint32_t {0x7FFFU + (uint32_t {bits >> 16} & 1U)}}
>> 16);
}
};
static_assert(sizeof(bfloat16_t) == 2, "bfloat16_t must be 2 bytes");
void cvt_float_to_bfloat16(bfloat16_t *out, const float *inp, size_t nelems);
void cvt_bfloat16_to_float(float *out, const bfloat16_t *inp, size_t nelems);
// performs element-by-element sum of inp and add float arrays and stores
// result to bfloat16 out array with downconversion
// out[:] = (bfloat16_t)(inp0[:] + inp1[:])
void add_floats_and_cvt_to_bfloat16(
bfloat16_t *out, const float *inp0, const float *inp1, size_t nelems);
} // namespace impl
} // namespace dnnl
#endif