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
// ----------------------------------------------------------------------------
// Copyright 2017 Nervana Systems Inc.
// 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
// ----------------------------------------------------------------------------
#include <iostream>
#include <vector>
#include <mkldnn.hpp>
#include "gtest/gtest.h"
static int tensor_volume(const mkldnn::memory::dims &t)
{
int x = 1;
for (const auto i : t)
x *= i;
return x;
}
TEST(mkldnn, engine)
{
using namespace mkldnn;
#pragma GCC diagnostic ignored "-Wgnu-statement-expression"
EXPECT_NO_THROW(({
auto cpu_engine = engine(engine::cpu, 0);
const int mb = 2;
const int groups = 2;
memory::dims input_tz = {mb, 256, 13, 13};
memory::dims weights_tz = {groups, 384/groups, 256/groups, 3, 3};
memory::dims bias_tz = {384};
memory::dims strides = {1, 1};
memory::dims padding = {0, 0};
memory::dims output_tz = {mb, 384,
(input_tz[2] + 2*padding[0] - weights_tz[3])/strides[0] + 1,
(input_tz[3] + 2*padding[1] - weights_tz[4])/strides[1] + 1,
};
std::vector<float> input(tensor_volume(input_tz), .0f);
std::vector<float> weights(tensor_volume(weights_tz), .0f);
std::vector<float> bias(tensor_volume(bias_tz), .0f);
std::vector<float> output(tensor_volume(output_tz), .0f);
auto c3_src_desc = memory::desc({input_tz}, memory::data_type::f32, memory::format::nchw);
auto c3_weights_desc = memory::desc({weights_tz}, memory::data_type::f32, memory::format::goihw);
auto c3_bias_desc = memory::desc({bias_tz}, memory::data_type::f32, memory::format::x);
auto c3_dst_desc = memory::desc({output_tz}, memory::data_type::f32, memory::format::nchw);
auto c3_src = memory({c3_src_desc, cpu_engine}, input.data());
auto c3_weights = memory({c3_weights_desc, cpu_engine}, weights.data());
auto c3_bias = memory({c3_bias_desc, cpu_engine}, bias.data());
auto c3_dst = memory({c3_dst_desc, cpu_engine}, output.data());
auto c3 = convolution_forward(convolution_forward::primitive_desc(convolution_forward::desc(prop_kind::forward,
algorithm::convolution_direct,
c3_src_desc, c3_weights_desc, c3_bias_desc, c3_dst_desc,
strides, padding, padding, padding_kind::zero),
cpu_engine), c3_src, c3_weights, c3_bias, c3_dst);
stream(stream::kind::eager).submit({c3}).wait();
}));
}