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
|
/*
* Copyright (c) 2018 Samsung Electronics Co., Ltd. All Rights Reserved
* Copyright (c) 2016-2018 ARM Limited.
*
* 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.
*/
#include "helpers.h"
#ifndef VEC_SIZE
#define VEC_SIZE 1
#endif
#if defined(OP_CODE) && defined(DATA_TYPE)
/** returns truth value of the two input tensors for BINARY LOGICAL OP.
* where BINARY LOGICAL OP can be AND, OR.
*
* @attention Data type can be passed using the -DDATA_TYPE compile flag, e.g. -DDATA_TYPE=uchar
* @attention Vector size should be given as a preprocessor argument using -DVEC_SIZE=size.
* e.g. -DVEC_SIZE=16
* @attention Operation type(code) specifying which operation to perform should be passed as
* preprocessor argument using -DOP_CODE = number. e.g. -DOP_CODE=1
*
* @param[in] input1_ptr Pointer to the source tensor.
* Supported data types: QASYMM8
* @param[in] input1_stride_x Stride of the source tensor in X dimension
* (in bytes)
* @param[in] input1_step_x input1_stride_x * number of elements along X
* processed per workitem(in bytes)
* @param[in] input1_stride_y Stride of the source tensor in Y dimension
* (in bytes)
* @param[in] input1_step_y input1_stride_y * number of elements along Y
* processed per workitem(in bytes)
* @param[in] input1_stride_z Stride of the source tensor in Z dimension
* (in bytes)
* @param[in] input1_step_z input1_stride_z * number of elements along Z
* processed per workitem(in bytes)
* @param[in] input1_offset_first_element_in_bytes The offset of the first element in the source
* tensor
* @param[in] input2_ptr Pointer to the source tensor.
* Supported data types: QASYMM8
* @param[in] input2_stride_x Stride of the source tensor in X dimension
* (in bytes)
* @param[in] input2_step_x input2_stride_x * number of elements along X
* processed per workitem(in bytes)
* @param[in] input2_stride_y Stride of the source tensor in Y dimension
* (in bytes)
* @param[in] input2_step_y input2_stride_y * number of elements along Y
* processed per workitem(in bytes)
* @param[in] input2_stride_z Stride of the source tensor in Z dimension
* (in bytes)
* @param[in] input2_step_z input2_stride_z * number of elements along Z
* processed per workitem(in bytes)
* @param[in] input2_offset_first_element_in_bytes The offset of the first element in the source
* tensor
* @param[out] output_ptr Pointer to the destination tensor.
* Supported data types: QASYMM8
* @param[in] output_stride_x Stride of the destination tensor in X dimension
* (in bytes)
* @param[in] output_step_x output_stride_x * number of elements along X
* processed per workitem(in bytes)
* @param[in] output_stride_y Stride of the destination tensor in Y dimension
* (in bytes)
* @param[in] output_step_y output_stride_y * number of elements along Y
* processed per workitem(in bytes)
* @param[in] output_stride_z Stride of the destination tensor in Z dimension
* (in bytes)
* @param[in] output_step_z output_stride_z * number of elements along Z
* processed per workitem(in bytes)
*/
__kernel void binary_logical_op(TENSOR3D_DECLARATION(input1), TENSOR3D_DECLARATION(input2),
TENSOR3D_DECLARATION(output))
{
Tensor3D input1 = CONVERT_TO_TENSOR3D_STRUCT(input1);
Tensor3D input2 = CONVERT_TO_TENSOR3D_STRUCT(input2);
Tensor3D output = CONVERT_TO_TENSOR3D_STRUCT(output);
#if OP_CODE == 1 // LOGICAL AND
VSTORE(VEC_SIZE)
(CONVERT(VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)input1.ptr) &&
VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)input2.ptr),
VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE)),
0, (__global DATA_TYPE *)output.ptr);
#elif OP_CODE == 2 // LOGICAL OR
VSTORE(VEC_SIZE)
(CONVERT(VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)input1.ptr) ||
VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)input2.ptr),
VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE)),
0, (__global DATA_TYPE *)output.ptr);
#else // OP NOT SUPPORTED
return
#endif
}
#endif // if defined(OP_CODE) && defined(DATA_TYPE)
|