aboutsummaryrefslogtreecommitdiff
path: root/src/core/CL
diff options
context:
space:
mode:
authorGeorgios Pinitas <georgios.pinitas@arm.com>2021-01-15 09:29:50 +0000
committerMichele Di Giorgio <michele.digiorgio@arm.com>2021-01-21 13:42:41 +0000
commitf47f718e87889b7d501a48e1d86b5d12b2e3f54c (patch)
tree6d55334ce1d4c54612320b53a59fee8e429b590b /src/core/CL
parentae182f24b20ca78342408dfbae7e66773b35f65f (diff)
downloadComputeLibrary-f47f718e87889b7d501a48e1d86b5d12b2e3f54c.tar.gz
Make CLFloor and CLActivation kernels and functions state-less
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Icbe4e6a7c6732a59bdda0136af44c4852452dfd1 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4891 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL')
-rw-r--r--src/core/CL/CLKernels.h2
-rw-r--r--src/core/CL/kernels/CLActivationLayerKernel.cpp250
-rw-r--r--src/core/CL/kernels/CLActivationLayerKernel.h77
-rw-r--r--src/core/CL/kernels/CLFloorKernel.cpp122
-rw-r--r--src/core/CL/kernels/CLFloorKernel.h81
5 files changed, 0 insertions, 532 deletions
diff --git a/src/core/CL/CLKernels.h b/src/core/CL/CLKernels.h
index 11f1d2d7cf..76bc5271b4 100644
--- a/src/core/CL/CLKernels.h
+++ b/src/core/CL/CLKernels.h
@@ -27,7 +27,6 @@
/* Header regrouping all the CL kernels */
#include "src/core/CL/kernels/CLAbsoluteDifferenceKernel.h"
#include "src/core/CL/kernels/CLAccumulateKernel.h"
-#include "src/core/CL/kernels/CLActivationLayerKernel.h"
#include "src/core/CL/kernels/CLArgMinMaxLayerKernel.h"
#include "src/core/CL/kernels/CLBatchNormalizationLayerKernel.h"
#include "src/core/CL/kernels/CLBatchToSpaceLayerKernel.h"
@@ -65,7 +64,6 @@
#include "src/core/CL/kernels/CLFFTScaleKernel.h"
#include "src/core/CL/kernels/CLFastCornersKernel.h"
#include "src/core/CL/kernels/CLFillBorderKernel.h"
-#include "src/core/CL/kernels/CLFloorKernel.h"
#include "src/core/CL/kernels/CLFuseBatchNormalizationKernel.h"
#include "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyNativeKernel.h"
#include "src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.h"
diff --git a/src/core/CL/kernels/CLActivationLayerKernel.cpp b/src/core/CL/kernels/CLActivationLayerKernel.cpp
deleted file mode 100644
index 00b6829e91..0000000000
--- a/src/core/CL/kernels/CLActivationLayerKernel.cpp
+++ /dev/null
@@ -1,250 +0,0 @@
-/*
- * Copyright (c) 2016-2020 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#include "src/core/CL/kernels/CLActivationLayerKernel.h"
-
-#include "arm_compute/core/CL/CLCoreRuntimeContext.h"
-#include "arm_compute/core/CL/CLHelpers.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/TensorInfo.h"
-#include "arm_compute/core/Utils.h"
-#include "src/core/CL/CLValidate.h"
-#include "src/core/helpers/AutoConfiguration.h"
-#include "src/core/helpers/WindowHelpers.h"
-#include "support/Cast.h"
-
-#include "support/StringSupport.h"
-
-#include <set>
-
-namespace arm_compute
-{
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, const ActivationLayerInfo &act_info)
-{
- ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
- ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::QASYMM8_SIGNED, DataType::QSYMM16, DataType::F16, DataType::F32);
-
- static std::set<ActivationLayerInfo::ActivationFunction> quantized_supported_activations =
- {
- ActivationLayerInfo::ActivationFunction::RELU,
- ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU,
- ActivationLayerInfo::ActivationFunction::BOUNDED_RELU,
- ActivationLayerInfo::ActivationFunction::LOGISTIC,
- ActivationLayerInfo::ActivationFunction::TANH,
- ActivationLayerInfo::ActivationFunction::HARD_SWISH,
- ActivationLayerInfo::ActivationFunction::LEAKY_RELU,
- };
- const DataType data_type = input->data_type();
- const QuantizationInfo &oq_info = (output != nullptr) ? output->quantization_info() : input->quantization_info();
- const ActivationLayerInfo::ActivationFunction f_act = act_info.activation();
-
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(is_data_type_quantized(data_type) && (quantized_supported_activations.count(f_act) == 0),
- "For Quantized data type only hard swish, leaky relu, tanh, logistic, relu and lower/upper bounded relu are supported");
-
- ARM_COMPUTE_RETURN_ERROR_ON(data_type == DataType::QASYMM8 && (f_act == ActivationLayerInfo::ActivationFunction::TANH) && (oq_info != QuantizationInfo(1.f / 128.f, 128)));
- ARM_COMPUTE_RETURN_ERROR_ON(data_type == DataType::QASYMM8 && (f_act == ActivationLayerInfo::ActivationFunction::LOGISTIC) && (oq_info != QuantizationInfo(1.f / 256.f, 0)));
-
- ARM_COMPUTE_RETURN_ERROR_ON(is_data_type_quantized_symmetric(data_type) && (f_act == ActivationLayerInfo::ActivationFunction::TANH) && (oq_info != QuantizationInfo(1.f / 32768.f, 0)));
- ARM_COMPUTE_RETURN_ERROR_ON(is_data_type_quantized_symmetric(data_type) && (f_act == ActivationLayerInfo::ActivationFunction::LOGISTIC) && (oq_info != QuantizationInfo(1.f / 32768.f, 0)));
-
- ARM_COMPUTE_RETURN_ERROR_ON(data_type == DataType::QASYMM8_SIGNED && (f_act == ActivationLayerInfo::ActivationFunction::TANH) && (oq_info != QuantizationInfo(1.f / 128.f, 0)));
- ARM_COMPUTE_RETURN_ERROR_ON(data_type == DataType::QASYMM8_SIGNED && (f_act == ActivationLayerInfo::ActivationFunction::LOGISTIC) && (oq_info != QuantizationInfo(1.f / 256.f, -128)));
-
- // Checks performed when output is configured
- if((output != nullptr) && (output->total_size() != 0))
- {
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(input, output);
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
- }
-
- return Status{};
-}
-} // namespace
-
-CLActivationLayerKernel::CLActivationLayerKernel()
- : _run_in_place(false)
-{
-}
-
-void CLActivationLayerKernel::configure(const CLCompileContext &compile_context, ITensorInfo *input, ITensorInfo *output, ActivationLayerInfo act_info)
-{
- ARM_COMPUTE_ERROR_ON_NULLPTR(input);
-
- auto padding_info = get_padding_info({ input, output });
-
- _run_in_place = (output == nullptr) || (output == input);
-
- if(output != nullptr)
- {
- // Output auto inizialitation if not yet initialized
- auto_init_if_empty(*output, *input->clone());
- }
-
- ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input, (output != nullptr) ? output : nullptr, act_info));
-
- const unsigned int num_elems_processed_per_iteration = adjust_vec_size(16 / input->element_size(), input->dimension(0));
-
- const DataType dt = input->data_type();
- float a_const = act_info.a();
- float b_const = act_info.b();
-
- const ActivationLayerInfo::ActivationFunction f_act = act_info.activation();
- const bool is_quantized = is_data_type_quantized(dt);
- const bool perform_activation_in_float =
- (f_act == ActivationLayerInfo::ActivationFunction::LOGISTIC)
- || (f_act == ActivationLayerInfo::ActivationFunction::TANH)
- || (f_act == ActivationLayerInfo::ActivationFunction::HARD_SWISH)
- || (f_act == ActivationLayerInfo::ActivationFunction::LEAKY_RELU);
-
- // Set build options
- CLBuildOptions build_opts;
- build_opts.add_option_if(perform_activation_in_float, "-DFLOAT_DOMAIN");
- build_opts.add_option_if(_run_in_place, "-DIN_PLACE");
- build_opts.add_option("-DACT=" + lower_string(string_from_activation_func(f_act)));
- build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(dt));
- build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_processed_per_iteration));
- build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + support::cpp11::to_string(input->dimension(0) % num_elems_processed_per_iteration));
-
- std::string kernel_name = std::string("activation_layer");
-
- // Set quantization info build options
- if(is_quantized)
- {
- const UniformQuantizationInfo iq_info = input->quantization_info().uniform();
-
- if(!perform_activation_in_float)
- {
- int a_const_int = 0;
- int b_const_int = 0;
-
- // Create quantized version of constants a, b if needed
- switch(dt)
- {
- case DataType::QASYMM8:
- {
- a_const_int = quantize_qasymm8(a_const, iq_info);
- b_const_int = quantize_qasymm8(b_const, iq_info);
- }
- break;
- case DataType::QASYMM8_SIGNED:
- {
- a_const_int = quantize_qasymm8_signed(a_const, iq_info);
- b_const_int = quantize_qasymm8_signed(b_const, iq_info);
- }
- break;
- case DataType::QSYMM16:
- {
- a_const_int = quantize_qsymm16(a_const, iq_info);
- b_const_int = quantize_qsymm16(b_const, iq_info);
- }
- break;
- default:
- break;
- }
- build_opts.add_option(("-DA_VAL=" + support::cpp11::to_string(a_const_int)));
- build_opts.add_option(("-DB_VAL=" + support::cpp11::to_string(b_const_int)));
- }
- else
- {
- build_opts.add_option(("-DA_VAL=" + float_to_string_with_full_precision(a_const)));
- build_opts.add_option(("-DB_VAL=" + float_to_string_with_full_precision(b_const)));
- }
-
- // Quantized value of 0 corresponds to the offset o1
- build_opts.add_option(("-DCONST_0=" + (is_data_type_quantized_asymmetric(dt) ? support::cpp11::to_string(iq_info.offset) : "0")));
- build_opts.add_option(("-DS1_VAL=" + float_to_string_with_full_precision(iq_info.scale)));
- build_opts.add_option_if(is_data_type_quantized_asymmetric(dt), "-DO1_VAL=" + support::cpp11::to_string(iq_info.offset));
-
- // Set correct kernel name
- kernel_name += perform_activation_in_float ? std::string("_quant_f32") : std::string("_quant");
-
- // Set scale and offset of the input and output if they have different quantization info
- if(output != nullptr)
- {
- const UniformQuantizationInfo oq_info = output->quantization_info().uniform();
-
- if(iq_info != oq_info)
- {
- build_opts.add_option(("-DS2_VAL=" + float_to_string_with_full_precision(oq_info.scale)));
- build_opts.add_option_if(is_data_type_quantized_asymmetric(dt), "-DO2_VAL=" + support::cpp11::to_string(oq_info.offset));
- }
- }
- }
- else
- {
- // Set A, B constants in build options for float types
- build_opts.add_option(("-DA_VAL=" + float_to_string_with_full_precision(a_const)));
- build_opts.add_option(("-DB_VAL=" + float_to_string_with_full_precision(b_const)));
- }
-
- // Create kernel
- _kernel = create_kernel(compile_context, kernel_name, build_opts.options());
-
- // Configure kernel window
- Window win = calculate_max_window(*input, Steps(num_elems_processed_per_iteration));
- ICLKernel::configure_internal(win);
-
- // Set config_id for enabling LWS tuning
- _config_id = "activation_layer_";
- _config_id += lower_string(string_from_data_type(dt));
- _config_id += "_";
- _config_id += support::cpp11::to_string(input->dimension(0));
- _config_id += "_";
- _config_id += support::cpp11::to_string(input->dimension(1));
-
- ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info));
-}
-
-Status CLActivationLayerKernel::validate(const ITensorInfo *input, const ITensorInfo *output, const ActivationLayerInfo &act_info)
-{
- ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, act_info));
- return Status{};
-}
-
-void CLActivationLayerKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
-{
- ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
- ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
-
- const auto src = utils::cast::polymorphic_downcast<const ICLTensor *>(tensors.get_const_tensor(TensorType::ACL_SRC));
- auto dst = utils::cast::polymorphic_downcast<ICLTensor *>(tensors.get_tensor(TensorType::ACL_DST));
- ARM_COMPUTE_ERROR_ON(_run_in_place && src != dst);
-
- Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
- Window slice = collapsed.first_slice_window_3D();
-
- do
- {
- unsigned int idx = 0;
- add_3D_tensor_argument(idx, src, slice);
- if(!_run_in_place)
- {
- add_3D_tensor_argument(idx, dst, slice);
- }
- enqueue(queue, *this, slice, lws_hint());
- }
- while(collapsed.slide_window_slice_3D(slice));
-}
-} // namespace arm_compute
diff --git a/src/core/CL/kernels/CLActivationLayerKernel.h b/src/core/CL/kernels/CLActivationLayerKernel.h
deleted file mode 100644
index 821418f835..0000000000
--- a/src/core/CL/kernels/CLActivationLayerKernel.h
+++ /dev/null
@@ -1,77 +0,0 @@
-/*
- * Copyright (c) 2016-2020 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#ifndef ARM_COMPUTE_CLACTIVATIONLAYERKERNEL_H
-#define ARM_COMPUTE_CLACTIVATIONLAYERKERNEL_H
-
-#include "src/core/CL/ICLKernel.h"
-
-namespace arm_compute
-{
-class ICLTensor;
-/** Interface for the activation layer kernel. */
-class CLActivationLayerKernel : public ICLKernel
-{
-public:
- /** Default constructor */
- CLActivationLayerKernel();
- /** Prevent instances of this class from being copied (As this class contains pointers) */
- CLActivationLayerKernel(const CLActivationLayerKernel &) = delete;
- /** Prevent instances of this class from being copied (As this class contains pointers) */
- CLActivationLayerKernel &operator=(const CLActivationLayerKernel &) = delete;
- /** Allow instances of this class to be moved */
- CLActivationLayerKernel(CLActivationLayerKernel &&) = default;
- /** Allow instances of this class to be moved */
- CLActivationLayerKernel &operator=(CLActivationLayerKernel &&) = default;
- /** Default destructor */
- ~CLActivationLayerKernel() = default;
- /** Set the input and output tensor.
- *
- * @note If the output tensor is a nullptr, the activation function will be performed in-place
- *
- * @param[in] compile_context The compile context to be used.
- * @param[in, out] input Source tensor. In case of @p output tensor = nullptr, this tensor will store the result
- * of the activation function. Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM16/F16/F32.
- * @param[out] output Destination tensor. Data type supported: same as @p input
- * @param[in] act_info Activation layer information.
- */
- void configure(const CLCompileContext &compile_context, ITensorInfo *input, ITensorInfo *output, ActivationLayerInfo act_info);
- /** Static function to check if given info will lead to a valid configuration of @ref CLActivationLayerKernel
- *
- * @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result
- * of the activation function. Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM16/F16/F32.
- * @param[in] output Destination tensor info. Data type supported: same as @p input
- * @param[in] act_info Activation layer information.
- *
- * @return a status
- */
- static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ActivationLayerInfo &act_info);
-
- // Inherited methods overridden:
- void run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue) override;
-
-private:
- bool _run_in_place;
-};
-} // namespace arm_compute
-#endif /*ARM_COMPUTE_CLACTIVATIONLAYERKERNEL_H */
diff --git a/src/core/CL/kernels/CLFloorKernel.cpp b/src/core/CL/kernels/CLFloorKernel.cpp
deleted file mode 100644
index 6af1822656..0000000000
--- a/src/core/CL/kernels/CLFloorKernel.cpp
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * Copyright (c) 2017-2021 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#include "src/core/CL/kernels/CLFloorKernel.h"
-
-#include "arm_compute/core/CL/CLHelpers.h"
-#include "arm_compute/core/CL/CLKernelLibrary.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/TensorInfo.h"
-#include "arm_compute/core/Utils.h"
-#include "arm_compute/core/Validate.h"
-#include "src/core/CL/CLValidate.h"
-#include "src/core/helpers/AutoConfiguration.h"
-#include "src/core/helpers/WindowHelpers.h"
-#include "support/StringSupport.h"
-
-namespace arm_compute
-{
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output)
-{
- ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
- ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
- ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F16, DataType::F32);
-
- // Validate in case of configured output
- if(output->total_size() > 0)
- {
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(input, output);
- }
-
- return Status{};
-}
-} // namespace
-
-CLFloorKernel::CLFloorKernel()
- : _input(nullptr), _output(nullptr)
-{
-}
-
-void CLFloorKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output)
-{
- ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
-
- // Auto initialize output
- auto_init_if_empty(*output->info(), input->info()->tensor_shape(), 1, input->info()->data_type());
-
- // Validate
- ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info()));
- auto padding_info = get_padding_info({ input, output });
-
- _input = input;
- _output = output;
-
- const unsigned int vec_size_x = adjust_vec_size(max_cl_vector_width / input->info()->element_size(), input->info()->dimension(0));
- const int vec_size_x_leftovers = input->info()->dimension(0) % vec_size_x;
- CLBuildOptions build_opts;
- build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(input->info()->data_type()));
- build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(vec_size_x));
- build_opts.add_option("-DVEC_SIZE_LEFTOVER=" + support::cpp11::to_string(vec_size_x_leftovers));
-
- // Create kernel
- _kernel = create_kernel(compile_context, "floor_layer", build_opts.options());
-
- // Configure kernel window
- Window win = calculate_max_window(*input->info(), Steps(vec_size_x));
- ICLKernel::configure_internal(win);
- ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info));
-}
-
-void CLFloorKernel::configure(const ICLTensor *input, ICLTensor *output)
-{
- configure(CLKernelLibrary::get().get_compile_context(), input, output);
-}
-
-Status CLFloorKernel::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
- ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output));
- return Status{};
-}
-
-void CLFloorKernel::run(const Window &window, cl::CommandQueue &queue)
-{
- ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
- ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
-
- Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
- Window slice = collapsed.first_slice_window_3D();
-
- do
- {
- unsigned int idx = 0;
- add_3D_tensor_argument(idx, _input, slice);
- add_3D_tensor_argument(idx, _output, slice);
- enqueue(queue, *this, slice, lws_hint());
- }
- while(collapsed.slide_window_slice_3D(slice));
-}
-} // namespace arm_compute
diff --git a/src/core/CL/kernels/CLFloorKernel.h b/src/core/CL/kernels/CLFloorKernel.h
deleted file mode 100644
index f5635141e4..0000000000
--- a/src/core/CL/kernels/CLFloorKernel.h
+++ /dev/null
@@ -1,81 +0,0 @@
-/*
- * Copyright (c) 2017-2020 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#ifndef ARM_COMPUTE_CLFLOORKERNEL_H
-#define ARM_COMPUTE_CLFLOORKERNEL_H
-
-#include "src/core/CL/ICLKernel.h"
-
-namespace arm_compute
-{
-class ICLTensor;
-
-/** OpenCL kernel to perform a floor operation */
-class CLFloorKernel : public ICLKernel
-{
-public:
- /** Default constructor */
- CLFloorKernel();
- /** Prevent instances of this class from being copied (As this class contains pointers) */
- CLFloorKernel(const CLFloorKernel &) = delete;
- /** Prevent instances of this class from being copied (As this class contains pointers) */
- CLFloorKernel &operator=(const CLFloorKernel &) = delete;
- /** Allow instances of this class to be moved */
- CLFloorKernel(CLFloorKernel &&) = default;
- /** Allow instances of this class to be moved */
- CLFloorKernel &operator=(CLFloorKernel &&) = default;
- /** Default destructor */
- ~CLFloorKernel() = default;
- /** Set the source, destination of the kernel
- *
- * @param[in] input Source tensor. Data type supported: F16/F32.
- * @param[out] output Destination tensor. Same as @p input
- */
- void configure(const ICLTensor *input, ICLTensor *output);
-
- /** Set the source, destination of the kernel
- *
- * @param[in] compile_context The compile context to be used.
- * @param[in] input Source tensor. Data type supported: F16/F32.
- * @param[out] output Destination tensor. Same as @p input
- */
- void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output);
-
- /** Static function to check if given info will lead to a valid configuration of @ref CLFloorKernel
- *
- * @param[in] input Source tensor info. Data type supported: F16/F32.
- * @param[in] output Destination tensor info. Same as @p input
- *
- * @return a status
- */
- static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-
- // Inherited methods overridden:
- void run(const Window &window, cl::CommandQueue &queue) override;
-
-private:
- const ICLTensor *_input;
- ICLTensor *_output;
-};
-} // namespace arm_compute
-#endif /*ARM_COMPUTE_CLFLOORKERNEL_H */