Compute Library
 20.08
NEL2NormalizeLayer.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2017-2019 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
25 
28 
29 namespace arm_compute
30 {
31 namespace
32 {
33 constexpr int max_input_tensor_dim = 3;
34 } // namespace
35 
36 NEL2NormalizeLayer::NEL2NormalizeLayer(std::shared_ptr<IMemoryManager> memory_manager)
37  : _memory_group(std::move(memory_manager)), _reduce_func(), _normalize_kernel(), _sumsq()
38 {
39 }
40 
42 {
43  // Manage intermediate buffers
44  _memory_group.manage(&_sumsq);
45 
46  // Configure Kernels
47  const uint32_t actual_axis = wrap_around(axis, max_input_tensor_dim);
48  _reduce_func.configure(input, &_sumsq, actual_axis, ReductionOperation::SUM_SQUARE);
49  _normalize_kernel.configure(input, &_sumsq, output, axis, epsilon);
50 
51  // Allocate intermediate tensors
52  _sumsq.allocator()->allocate();
53 }
54 
56 {
57  TensorShape shape(input->tensor_shape());
58 
59  // Create intermediate tensor info
60  TensorInfo sum_sq;
61  sum_sq.set_data_type(input->data_type());
62  sum_sq.set_tensor_shape(shape);
63 
64  const uint32_t actual_axis = wrap_around(axis, max_input_tensor_dim);
66 
67  // Reduce shape on axis
68  shape.set(actual_axis, 1);
69  sum_sq.set_tensor_shape(shape);
70 
72 
73  return Status{};
74 }
75 
77 {
78  MemoryGroupResourceScope scope_mg(_memory_group);
79 
80  _reduce_func.run();
81  NEScheduler::get().schedule(&_normalize_kernel, Window::DimY);
82 }
83 } // namespace arm_compute
Shape of a tensor.
Definition: TensorShape.h:39
void run() override
Run the kernels contained in the function.
virtual ITensorInfo & set_tensor_shape(const TensorShape &shape)=0
Set the shape of an already initialized tensor.
void run() override
Run the kernels contained in the function.
#define ARM_COMPUTE_RETURN_ON_ERROR(status)
Checks if a status contains an error and returns it.
Definition: Error.h:204
ITensorInfo & set_data_type(DataType data_type) override
Set the data type to the specified value.
Definition: TensorInfo.cpp:319
Store the tensor's metadata.
Definition: ITensorInfo.h:40
Status class.
Definition: Error.h:52
void configure(ITensor *input, ITensor *output, int axis, float epsilon=1e-12f)
Set the input and output tensors.
Interface for NEON tensor.
Definition: ITensor.h:36
Copyright (c) 2017-2020 Arm Limited.
TensorAllocator * allocator()
Return a pointer to the tensor's allocator.
Definition: Tensor.cpp:48
void manage(IMemoryManageable *obj) override
Sets a object to be managed by the given memory group.
Definition: MemoryGroup.h:79
T wrap_around(T x, T m)
Wrap-around a number within the range 0 <= x < m.
Definition: Helpers.h:799
static Status validate(const ITensorInfo *input, const ITensorInfo *output, unsigned int axis, ReductionOperation op, bool keep_dims=true)
Static function to check if given info will lead to a valid configuration of NEReductionOperation.
void allocate() override
Allocate size specified by TensorInfo of CPU memory.
static Status validate(const ITensorInfo *input, const ITensorInfo *output, int axis, float epsilon=1e-12f)
Static function to check if given info will lead to a valid configuration of NEL2NormalizeLayer.
NEL2NormalizeLayer(std::shared_ptr< IMemoryManager > memory_manager=nullptr)
Constructor.
static constexpr size_t DimY
Alias for dimension 1 also known as Y dimension.
Definition: Window.h:45
static Status validate(const ITensorInfo *input, const ITensorInfo *sum, const ITensorInfo *output, int axis, float epsilon)
Static function to check if given info will lead to a valid configuration of NEL2NormalizeLayerKernel...
Memory group resources scope handling class.
Definition: IMemoryGroup.h:82
virtual void schedule(ICPPKernel *kernel, const Hints &hints)=0
Runs the kernel in the same thread as the caller synchronously.
Store the tensor's metadata.
Definition: TensorInfo.h:45
void configure(ITensor *input, ITensor *output, unsigned int axis, ReductionOperation op, bool keep_dims=true)
Set the input and output tensors.
static IScheduler & get()
Access the scheduler singleton.
Definition: Scheduler.cpp:95
void configure(const ITensor *input, const ITensor *sum, ITensor *output, int axis, float epsilon)
Set the input and output tensors.