Compute Library
 21.02
CLAccumulateKernel.cpp
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2016-2020 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 
29 #include "arm_compute/core/Error.h"
31 #include "arm_compute/core/Types.h"
33 
34 namespace arm_compute
35 {
36 namespace
37 {
38 constexpr unsigned int num_elems_processed_per_iteration = 16;
39 } // namespace
40 
42 {
43  configure(CLKernelLibrary::get().get_compile_context(), input, accum);
44 }
45 
46 void CLAccumulateKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *accum)
47 {
50 
51  // Create kernel
52  _kernel = create_kernel(compile_context, "accumulate");
53 
54  // Make sure _kernel is initialized before calling the parent's configure
56 }
57 
59 {
60  configure(CLKernelLibrary::get().get_compile_context(), input, alpha, accum);
61 }
62 
63 void CLAccumulateWeightedKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, float alpha, ICLTensor *accum)
64 {
67  ARM_COMPUTE_ERROR_ON(alpha < 0.0 || alpha > 1.0);
68 
69  // Create kernel
70  _kernel = create_kernel(compile_context, "accumulate_weighted");
71 
72  // Set static kernel arguments
73  unsigned int idx = 2 * num_arguments_per_2D_tensor(); //Skip the input and output parameters
74  _kernel.setArg(idx++, alpha);
75 
76  // Configure kernel window
78 }
79 
80 void CLAccumulateSquaredKernel::configure(const ICLTensor *input, uint32_t shift, ICLTensor *accum)
81 {
82  configure(CLKernelLibrary::get().get_compile_context(), input, shift, accum);
83 }
84 
85 void CLAccumulateSquaredKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, uint32_t shift, ICLTensor *accum)
86 {
89  ARM_COMPUTE_ERROR_ON(shift > 15);
90 
91  // Create kernel
92  _kernel = create_kernel(compile_context, "accumulate_squared");
93 
94  // Set static kernel arguments
95  unsigned int idx = 2 * num_arguments_per_2D_tensor(); //Skip the input and output parameters
96  _kernel.setArg(idx++, shift);
97 
98  // Configure kernel window
100 }
101 } // namespace arm_compute
void configure(const ICLTensor *input, float alpha, ICLTensor *accum)
Set the input and accumulation images, and the scale value.
void configure(const ICLTensor *input, uint32_t shift, ICLTensor *accum)
Set the input and accumulation tensors and the shift value.
1 channel, 1 U8 per channel
#define ARM_COMPUTE_ERROR_ON(cond)
If the condition is true then an error message is printed and an exception thrown.
Definition: Error.h:466
static CLKernelLibrary & get()
Access the KernelLibrary singleton.
Copyright (c) 2017-2021 Arm Limited.
void configure(const ICLTensor *input, ICLTensor *output, unsigned int num_elems_processed_per_iteration, bool border_undefined=false, const BorderSize &border_size=BorderSize())
Configure the kernel.
cl::Kernel create_kernel(const CLCompileContext &ctx, const std::string &kernel_name, const std::set< std::string > &build_opts=std::set< std::string >())
Creates an opencl kernel using a compile context.
Definition: CLHelpers.cpp:403
static constexpr unsigned int num_arguments_per_2D_tensor()
Returns the number of arguments enqueued per 2D tensor object.
Definition: ICLKernel.h:206
1 channel, 1 S16 per channel
#define ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(t, c,...)
Definition: Validate.h:790
CLCompileContext class.
Interface for OpenCL tensor.
Definition: ICLTensor.h:42
Manages all the OpenCL kernels compilation and caching, provides accessors for the OpenCL Context...
Wrapper to configure the Khronos OpenCL C++ header.
unsigned int num_elems_processed_per_iteration
void configure(const ICLTensor *input, ICLTensor *accum)
Set the input and accumulation tensors.