forked from NVIDIA/cutlass
-
Notifications
You must be signed in to change notification settings - Fork 29
/
Copy pathtensor_silu.h
139 lines (109 loc) · 4.44 KB
/
tensor_silu.h
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
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
/***************************************************************************************************
* Copyright (c) 2024 - 2025 Codeplay Software Ltd. All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice, this
* list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
*
* 3. Neither the name of the copyright holder nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
**************************************************************************************************/
#pragma once
// Cutlass includes
#include "cutlass/cutlass.h"
#include "cutlass/tensor_view.h"
#include "cutlass/util/reference/device/tensor_foreach.h"
///////////////////////////////////////////////////////////////////////////////////////////////////
namespace cutlass {
namespace reference {
namespace device {
///////////////////////////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////////////////////////
namespace detail {
template <
typename Element, ///< Element type
typename Layout> ///< Layout function
struct TensorSiLuFunc {
/// View type
using TensorView = TensorView<Element, Layout>;
/// Coordinate in tensor's index space
using TensorCoord = typename TensorView::TensorCoord;
/// Parameters structure
struct Params {
//
// Data members
//
TensorView view_in0, view_in1, view_out;
//
// Methods
//
Params(
TensorView view_out_ = TensorView(),
TensorView view_in0_ = TensorView(),
TensorView view_in1_ = TensorView()
):
view_out(view_out_), view_in0(view_in0_), view_in1(view_in1_){
}
};
//
// Data members
//
Params params;
//
// Methods
//
CUTLASS_DEVICE
TensorSiLuFunc(Params const ¶ms): params(params) {
}
CUTLASS_DEVICE
void operator()(TensorCoord const &coord) {
Element const& in0_val = params.view_in0.at(coord);
Element const& in1_val = params.view_in1.at(coord);
cutlass::epilogue::thread::SiLu<Element> silu;
cutlass::multiplies<Element> mul;
auto silu_lhs = silu(in0_val);
params.view_out.at(coord) = mul(silu_lhs, in1_val);
}
};
} // namespace detail
///////////////////////////////////////////////////////////////////////////////////////////////////
/// Apply SiLu on a tensor
template <
typename Element, ///< Element type
typename Layout> ///< Layout function
void TensorSiLu(
TensorView<Element, Layout> view_out, ///< destination tensor
TensorView<Element, Layout> view_in0, ///< source tensor
TensorView<Element, Layout> view_in1) { ///< source tensor
using Func = detail::TensorSiLuFunc<Element, Layout>;
using Params = typename Func::Params;
TensorForEach<Func, Layout::kRank, Params>(
view_out.extent(),
Params(view_out, view_in0, view_in1)
);
}
///////////////////////////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////////////////////////
} // namespace device
} // namespace reference
} // namespace cutlass