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
|
/*******************************************************************************
* Copyright 2020-2025 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.
*******************************************************************************/
/// @example sum.cpp
/// > Annotated version: @ref sum_example_cpp
///
/// @page sum_example_cpp_short
///
/// This C++ API example demonstrates how to create and execute a
/// [Sum](@ref dev_guide_sum) primitive.
///
/// Key optimizations included in this example:
/// - Identical memory formats for source (src) and destination (dst) tensors.
///
/// @page sum_example_cpp Sum Primitive Example
/// @copydetails sum_example_cpp_short
///
/// @include sum.cpp
#include <algorithm>
#include <cmath>
#include <iostream>
#include <string>
#include <vector>
#include "example_utils.hpp"
#include "oneapi/dnnl/dnnl.hpp"
using namespace dnnl;
void sum_example(dnnl::engine::kind engine_kind) {
// Create execution dnnl::engine.
dnnl::engine engine(engine_kind, 0);
// Create dnnl::stream.
dnnl::stream engine_stream(engine);
// Tensor dimensions.
const memory::dim N = 3, // batch size
IC = 3, // channels
IH = 227, // tensor height
IW = 227; // tensor width
// Source (src) and destination (dst) tensors dimensions.
memory::dims src_dims = {N, IC, IH, IW};
// Allocate buffers.
std::vector<float> src_data(product(src_dims));
std::vector<float> dst_data(product(src_dims));
// Initialize src.
std::generate(src_data.begin(), src_data.end(), []() {
static int i = 0;
return std::cos(i++ / 10.f);
});
// Number of src tensors.
const int num_src = 10;
// Scaling factors.
std::vector<float> scales(num_src);
std::generate(scales.begin(), scales.end(),
[](int n = 0) { return sin(float(n)); });
// Create an array of memory descriptors and memory objects for src tensors.
std::vector<memory::desc> src_md;
std::vector<memory> src_mem;
for (int n = 0; n < num_src; ++n) {
src_md.emplace_back(
src_dims, memory::data_type::f32, memory::format_tag::nchw);
src_mem.emplace_back(src_md.back(), engine);
// Write data to memory object's handle.
write_to_dnnl_memory(src_data.data(), src_mem.back());
}
// Create primitive descriptor.
auto sum_pd = sum::primitive_desc(engine, scales, src_md);
// Create the primitive.
auto sum_prim = sum(sum_pd);
// Create memory object for dst.
auto dst_mem = memory(sum_pd.dst_desc(), engine);
// Primitive arguments.
std::unordered_map<int, memory> sum_args;
sum_args.insert({DNNL_ARG_DST, dst_mem});
for (int n = 0; n < num_src; ++n) {
sum_args.insert({DNNL_ARG_MULTIPLE_SRC + n, src_mem[n]});
}
// Primitive execution: sum.
sum_prim.execute(engine_stream, sum_args);
// Wait for the computation to finalize.
engine_stream.wait();
// Read data from memory object's handle.
read_from_dnnl_memory(dst_data.data(), dst_mem);
}
int main(int argc, char **argv) {
return handle_example_errors(sum_example, parse_engine_kind(argc, argv));
}
|
