diff options
Diffstat (limited to 'arm_compute/runtime/CL')
99 files changed, 1354 insertions, 519 deletions
diff --git a/arm_compute/runtime/CL/CLArray.h b/arm_compute/runtime/CL/CLArray.h index 7efe208b9f..6e81a46a29 100644 --- a/arm_compute/runtime/CL/CLArray.h +++ b/arm_compute/runtime/CL/CLArray.h @@ -38,8 +38,7 @@ class CLArray : public ICLArray<T> { public: /** Default constructor: empty array */ - CLArray() - : ICLArray<T>(0), _buffer() + CLArray() : ICLArray<T>(0), _buffer() { } /** Prevent instances of this class from being copied (As this class contains pointers) */ @@ -55,7 +54,8 @@ public: * @param[in] max_num_values Maximum number of values the array will be able to stored */ CLArray(size_t max_num_values) - : ICLArray<T>(max_num_values), _buffer(CLScheduler::get().context(), CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_WRITE, max_num_values * sizeof(T)) + : ICLArray<T>(max_num_values), + _buffer(CLScheduler::get().context(), CL_MEM_ALLOC_HOST_PTR | CL_MEM_READ_WRITE, max_num_values * sizeof(T)) { } /** Enqueue a map operation of the allocated buffer. @@ -91,7 +91,8 @@ protected: uint8_t *do_map(cl::CommandQueue &q, bool blocking) override { ARM_COMPUTE_ERROR_ON(nullptr == _buffer.get()); - return static_cast<uint8_t *>(q.enqueueMapBuffer(_buffer, blocking ? CL_TRUE : CL_FALSE, CL_MAP_READ | CL_MAP_WRITE, 0, this->max_num_values() * sizeof(T))); + return static_cast<uint8_t *>(q.enqueueMapBuffer( + _buffer, blocking ? CL_TRUE : CL_FALSE, CL_MAP_READ | CL_MAP_WRITE, 0, this->max_num_values() * sizeof(T))); } void do_unmap(cl::CommandQueue &q, uint8_t *mapping) override { @@ -114,5 +115,5 @@ using CLInt16Array = CLArray<cl_short>; using CLInt32Array = CLArray<cl_int>; /** OpenCL Array of floats. */ using CLFloatArray = CLArray<cl_float>; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLARRAY_H */ diff --git a/arm_compute/runtime/CL/CLBufferAllocator.h b/arm_compute/runtime/CL/CLBufferAllocator.h index 7467e9d1c6..00ff017012 100644 --- a/arm_compute/runtime/CL/CLBufferAllocator.h +++ b/arm_compute/runtime/CL/CLBufferAllocator.h @@ -35,9 +35,9 @@ class CLBufferAllocator final : public IAllocator { public: // Inherited methods overridden: - void *allocate(size_t size, size_t alignment) override; - void free(void *ptr) override; + void *allocate(size_t size, size_t alignment) override; + void free(void *ptr) override; std::unique_ptr<IMemoryRegion> make_region(size_t size, size_t alignment) override; }; -} // arm_compute +} // namespace arm_compute #endif /*ARM_COMPUTE_CLBUFFERALLOCATOR_H */ diff --git a/arm_compute/runtime/CL/CLMemory.h b/arm_compute/runtime/CL/CLMemory.h index 7adee66c73..5abe86bd53 100644 --- a/arm_compute/runtime/CL/CLMemory.h +++ b/arm_compute/runtime/CL/CLMemory.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_RUNTIME_CL_CLMEMORY_H #define ARM_COMPUTE_RUNTIME_CL_CLMEMORY_H -#include "arm_compute/runtime/IMemory.h" - #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/runtime/CL/CLMemoryRegion.h" +#include "arm_compute/runtime/IMemory.h" #include <cstddef> #include <memory> @@ -75,8 +74,8 @@ public: // Inherited methods overridden: IMemoryRegion *region() final; IMemoryRegion *region() const final; - void set_region(IMemoryRegion *region) final; - void set_owned_region(std::unique_ptr<IMemoryRegion> region) final; + void set_region(IMemoryRegion *region) final; + void set_owned_region(std::unique_ptr<IMemoryRegion> region) final; private: ICLMemoryRegion *_region; diff --git a/arm_compute/runtime/CL/CLMemoryRegion.h b/arm_compute/runtime/CL/CLMemoryRegion.h index 66a30fa56b..365973a9e6 100644 --- a/arm_compute/runtime/CL/CLMemoryRegion.h +++ b/arm_compute/runtime/CL/CLMemoryRegion.h @@ -110,7 +110,7 @@ public: // Inherited methods overridden : void *ptr() final; void *map(cl::CommandQueue &q, bool blocking) final; - void unmap(cl::CommandQueue &q) final; + void unmap(cl::CommandQueue &q) final; }; /** OpenCL SVM memory region interface */ @@ -156,7 +156,7 @@ public: // Inherited methods overridden : void *map(cl::CommandQueue &q, bool blocking) final; - void unmap(cl::CommandQueue &q) final; + void unmap(cl::CommandQueue &q) final; }; /** OpenCL fine-grain SVM memory region implementation */ @@ -173,7 +173,7 @@ public: // Inherited methods overridden : void *map(cl::CommandQueue &q, bool blocking) final; - void unmap(cl::CommandQueue &q) final; + void unmap(cl::CommandQueue &q) final; }; } // namespace arm_compute #endif /* ARM_COMPUTE_RUNTIME_CL_CL_MEMORY_REGION_H */ diff --git a/arm_compute/runtime/CL/CLRuntimeContext.h b/arm_compute/runtime/CL/CLRuntimeContext.h index dd17645fa7..2ed4b74796 100644 --- a/arm_compute/runtime/CL/CLRuntimeContext.h +++ b/arm_compute/runtime/CL/CLRuntimeContext.h @@ -54,11 +54,11 @@ public: CLKernelLibrary &kernel_library(); private: - std::unique_ptr<CLScheduler> _gpu_owned_scheduler{ nullptr }; - CLScheduler *_gpu_scheduler{ nullptr }; - CLTuner _tuner{ false }; + std::unique_ptr<CLScheduler> _gpu_owned_scheduler{nullptr}; + CLScheduler *_gpu_scheduler{nullptr}; + CLTuner _tuner{false}; CLSymbols _symbols{}; - CLBackendType _backend_type{ CLBackendType::Native }; + CLBackendType _backend_type{CLBackendType::Native}; }; } // namespace arm_compute #endif /*ARM_COMPUTE_CLRUNTIME_CONTEXT_H */ diff --git a/arm_compute/runtime/CL/CLScheduler.h b/arm_compute/runtime/CL/CLScheduler.h index 3030239270..b74fcb74ef 100644 --- a/arm_compute/runtime/CL/CLScheduler.h +++ b/arm_compute/runtime/CL/CLScheduler.h @@ -28,8 +28,8 @@ #include "arm_compute/core/CL/CLTypes.h" #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/core/Error.h" -#include "arm_compute/core/Types.h" #include "arm_compute/core/experimental/Types.h" +#include "arm_compute/core/Types.h" #include "arm_compute/runtime/CL/CLGEMMHeuristicsHandle.h" #include "arm_compute/runtime/CL/CLHelpers.h" #include "arm_compute/runtime/CL/CLTypes.h" @@ -63,7 +63,9 @@ public: * @param[in] gemm_h (Optional) Pointer to CLGEMMHeuristicsHandle (default = nullptr) * @param[in] cl_backend_type (Optional) Type of backend to use (default = CLBackendType::Native) */ - void default_init(ICLTuner *cl_tuner = nullptr, CLGEMMHeuristicsHandle *gemm_h = nullptr, CLBackendType cl_backend_type = CLBackendType::Native); + void default_init(ICLTuner *cl_tuner = nullptr, + CLGEMMHeuristicsHandle *gemm_h = nullptr, + CLBackendType cl_backend_type = CLBackendType::Native); /** Initialises the scheduler with context and device provided by the user * * @param[in] device OpenCL device to be used @@ -71,7 +73,10 @@ public: * @param[in] cl_tuner (Optional) Pointer to ICLTuner (default=nullptr) * @param[in] gemm_h (Optional) Pointer to CLGEMMHeuristicsHandle (default = nullptr) */ - void default_init_with_context(cl::Device &device, cl::Context &ctx, ICLTuner *cl_tuner = nullptr, CLGEMMHeuristicsHandle *gemm_h = nullptr); + void default_init_with_context(cl::Device &device, + cl::Context &ctx, + ICLTuner *cl_tuner = nullptr, + CLGEMMHeuristicsHandle *gemm_h = nullptr); /** Re-initializes the context and command queue used by the scheduler to default values * and sets a default device and kernel path for the @ref CLKernelLibrary. @@ -80,7 +85,9 @@ public: * @param[in] gemm_h (Optional) Pointer to CLGEMMHeuristicsHandle (default = nullptr) * @param[in] cl_backend_type (Optional) Type of backend to use (default = CLBackendType::Native) */ - void default_reinit(ICLTuner *cl_tuner = nullptr, CLGEMMHeuristicsHandle *gemm_h = nullptr, CLBackendType cl_backend_type = CLBackendType::Native); + void default_reinit(ICLTuner *cl_tuner = nullptr, + CLGEMMHeuristicsHandle *gemm_h = nullptr, + CLBackendType cl_backend_type = CLBackendType::Native); /** Schedule the execution of the passed kernel if possible. * @@ -105,8 +112,12 @@ public: * @param[in] gemm_h (Optional) Pointer to CLGEMMHeuristicsHandle (default = nullptr) * @param[in] cl_backend_type (Optional) Type of backend to use (default = CLBackendType::Native) */ - void init(cl::Context context, cl::CommandQueue queue, const cl::Device &device, ICLTuner *cl_tuner = nullptr, CLGEMMHeuristicsHandle *gemm_h = nullptr, - CLBackendType cl_backend_type = CLBackendType::Native); + void init(cl::Context context, + cl::CommandQueue queue, + const cl::Device &device, + ICLTuner *cl_tuner = nullptr, + CLGEMMHeuristicsHandle *gemm_h = nullptr, + CLBackendType cl_backend_type = CLBackendType::Native); /** Accessor for the associated CL context. * diff --git a/arm_compute/runtime/CL/CLSubTensor.h b/arm_compute/runtime/CL/CLSubTensor.h index 0a7f5f89b2..c18df8086a 100644 --- a/arm_compute/runtime/CL/CLSubTensor.h +++ b/arm_compute/runtime/CL/CLSubTensor.h @@ -46,7 +46,10 @@ public: * @param[in] coords Coordinates of the first subtensor element inside the parent tensor. * @param[in] extend_parent (Optional) Extend parent with subtensor shape if subtensor indexes out of bounds */ - CLSubTensor(ICLTensor *parent, const TensorShape &tensor_shape, const Coordinates &coords, bool extend_parent = false); + CLSubTensor(ICLTensor *parent, + const TensorShape &tensor_shape, + const Coordinates &coords, + bool extend_parent = false); /** Destructor: free the tensor's memory */ ~CLSubTensor() = default; /** Restrict instances of this class to be copy constructed */ @@ -93,11 +96,11 @@ public: protected: // Inherited methods overridden: uint8_t *do_map(cl::CommandQueue &q, bool blocking) override; - void do_unmap(cl::CommandQueue &q) override; + void do_unmap(cl::CommandQueue &q) override; private: ICLTensor *_parent; mutable SubTensorInfo _info; }; -} +} // namespace arm_compute #endif /*ARM_COMPUTE_CLSUBTENSOR_H */ diff --git a/arm_compute/runtime/CL/CLTensor.h b/arm_compute/runtime/CL/CLTensor.h index ae73351f27..0729935e9e 100644 --- a/arm_compute/runtime/CL/CLTensor.h +++ b/arm_compute/runtime/CL/CLTensor.h @@ -87,17 +87,17 @@ public: TensorInfo *info() override; const cl::Buffer &cl_buffer() const override; CLQuantization quantization() const override; - void associate_memory_group(IMemoryGroup *memory_group) override; + void associate_memory_group(IMemoryGroup *memory_group) override; CLRuntimeContext *context(); protected: // Inherited methods overridden: uint8_t *do_map(cl::CommandQueue &q, bool blocking) override; - void do_unmap(cl::CommandQueue &q) override; + void do_unmap(cl::CommandQueue &q) override; private: mutable CLTensorAllocator _allocator; /**< Instance of the OpenCL tensor allocator */ - CLRuntimeContext *_ctx{ nullptr }; + CLRuntimeContext *_ctx{nullptr}; }; /** OpenCL Image */ diff --git a/arm_compute/runtime/CL/CLTensorAllocator.h b/arm_compute/runtime/CL/CLTensorAllocator.h index 1b061ee1d6..fde8e9c43a 100644 --- a/arm_compute/runtime/CL/CLTensorAllocator.h +++ b/arm_compute/runtime/CL/CLTensorAllocator.h @@ -24,15 +24,14 @@ #ifndef ARM_COMPUTE_CLTENSORALLOCATOR_H #define ARM_COMPUTE_CLTENSORALLOCATOR_H +#include "arm_compute/core/CL/CLTypes.h" +#include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/runtime/CL/CLArray.h" #include "arm_compute/runtime/CL/CLMemory.h" #include "arm_compute/runtime/IAllocator.h" #include "arm_compute/runtime/ITensorAllocator.h" #include "arm_compute/runtime/MemoryGroup.h" -#include "arm_compute/core/CL/CLTypes.h" -#include "arm_compute/core/CL/OpenCL.h" - #include <cstdint> namespace arm_compute @@ -148,7 +147,7 @@ private: static const cl::Buffer _empty_buffer; private: - CLRuntimeContext *_ctx; + CLRuntimeContext *_ctx; IMemoryManageable *_owner; /**< Memory manageable object that owns the allocator */ IMemoryGroup *_associated_memory_group; /**< Registered memory manager */ CLMemory _memory; /**< OpenCL memory */ diff --git a/arm_compute/runtime/CL/CLTuner.h b/arm_compute/runtime/CL/CLTuner.h index 93aa45adc1..cf293d3d27 100644 --- a/arm_compute/runtime/CL/CLTuner.h +++ b/arm_compute/runtime/CL/CLTuner.h @@ -153,9 +153,9 @@ private: std::unordered_map<std::string, CLTuningParams> _tuning_params_table; std::unordered_map<std::string, cl::NDRange> _lws_table; - cl::Event _kernel_event; - bool _tune_new_kernels; - CLTuningInfo _tuning_info; + cl::Event _kernel_event; + bool _tune_new_kernels; + CLTuningInfo _tuning_info; }; } // namespace arm_compute #endif /*ARM_COMPUTE_CLTUNER_H */ diff --git a/arm_compute/runtime/CL/CLTunerTypes.h b/arm_compute/runtime/CL/CLTunerTypes.h index 508cafac95..d9b914676a 100644 --- a/arm_compute/runtime/CL/CLTunerTypes.h +++ b/arm_compute/runtime/CL/CLTunerTypes.h @@ -43,7 +43,7 @@ enum class CLTunerMode struct CLTuningInfo { CLTunerMode tuner_mode = CLTunerMode::NORMAL; /**< Parameter to select the level (granularity) of the tuning */ - bool tune_wbsm = false; /**< Flag to tune the batches of work groups distributed to compute units. + bool tune_wbsm = false; /**< Flag to tune the batches of work groups distributed to compute units. Internally, the library will check if this feature is available on the target platform. This OpenCL tuner extension is still in experimental phase */ }; @@ -56,11 +56,10 @@ struct CLTuningInfo */ inline CLTunerMode tuner_mode_from_name(const std::string &name) { - static const std::map<std::string, CLTunerMode> tuner_modes = - { - { "exhaustive", CLTunerMode::EXHAUSTIVE }, - { "normal", CLTunerMode::NORMAL }, - { "rapid", CLTunerMode::RAPID }, + static const std::map<std::string, CLTunerMode> tuner_modes = { + {"exhaustive", CLTunerMode::EXHAUSTIVE}, + {"normal", CLTunerMode::NORMAL}, + {"rapid", CLTunerMode::RAPID}, }; #ifndef ARM_COMPUTE_EXCEPTIONS_DISABLED @@ -71,7 +70,7 @@ inline CLTunerMode tuner_mode_from_name(const std::string &name) #ifndef ARM_COMPUTE_EXCEPTIONS_DISABLED } - catch(const std::out_of_range &) + catch (const std::out_of_range &) { throw std::invalid_argument(name); } diff --git a/arm_compute/runtime/CL/CLTuningParams.h b/arm_compute/runtime/CL/CLTuningParams.h index 1e5ab25c26..a876fad112 100644 --- a/arm_compute/runtime/CL/CLTuningParams.h +++ b/arm_compute/runtime/CL/CLTuningParams.h @@ -26,6 +26,7 @@ #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/runtime/CL/CLTunerTypes.h" + #include "support/StringSupport.h" #include <ostream> @@ -36,8 +37,7 @@ namespace arm_compute class CLTuningParams { public: - CLTuningParams(const CLTuningParams &tuning_params) - : _lws(tuning_params._lws), _wbsm(tuning_params._wbsm) + CLTuningParams(const CLTuningParams &tuning_params) : _lws(tuning_params._lws), _wbsm(tuning_params._wbsm) { } @@ -45,18 +45,16 @@ public: : _lws(lws_x, lws_y, lws_z), _wbsm(wbsm) { } - CLTuningParams(cl::NDRange lws, cl_int wbsm = 0) - : _lws(lws), _wbsm(wbsm) + CLTuningParams(cl::NDRange lws, cl_int wbsm = 0) : _lws(lws), _wbsm(wbsm) { } - CLTuningParams(cl_int wbsm) - : CLTuningParams(cl::NullRange, wbsm) + CLTuningParams(cl_int wbsm) : CLTuningParams(cl::NullRange, wbsm) { } - CLTuningParams& operator=(const CLTuningParams &other) + CLTuningParams &operator=(const CLTuningParams &other) { - _lws = other._lws; + _lws = other._lws; _wbsm = other._wbsm; return *this; } @@ -84,8 +82,9 @@ public: std::string to_string(CLTuningInfo tuning_info) { std::string tuning_params_string = ""; - tuning_params_string += ";" + support::cpp11::to_string(_lws[0]) + ";" + support::cpp11::to_string(_lws[1]) + ";" + support::cpp11::to_string(_lws[2]); - if(tuning_info.tune_wbsm) + tuning_params_string += ";" + support::cpp11::to_string(_lws[0]) + ";" + support::cpp11::to_string(_lws[1]) + + ";" + support::cpp11::to_string(_lws[2]); + if (tuning_info.tune_wbsm) { tuning_params_string += ";" + support::cpp11::to_string(_wbsm); } @@ -98,19 +97,19 @@ public: std::vector<std::string> array; std::stringstream ss(tuning_params_string); std::string temp; - while(ss >> temp) + while (ss >> temp) { array.push_back(temp); } // Read 3 values for lws - if(array.size() < 3) + if (array.size() < 3) { return false; } const unsigned int lws_0 = support::cpp11::stoi(array[0]); const unsigned int lws_1 = support::cpp11::stoi(array[1]); const unsigned int lws_2 = support::cpp11::stoi(array[2]); - if(lws_0 == 0 && lws_1 == 0 && lws_2 == 0) + if (lws_0 == 0 && lws_1 == 0 && lws_2 == 0) { // If lws values are 0, cl::NullRange has to be used // otherwise the lws object will be badly created @@ -121,9 +120,9 @@ public: _lws = cl::NDRange(lws_0, lws_1, lws_2); } array.erase(array.begin(), array.begin() + 3); - if(tuning_info.tune_wbsm) + if (tuning_info.tune_wbsm) { - if(array.size() < 1) + if (array.size() < 1) { return false; } diff --git a/arm_compute/runtime/CL/CLTypes.h b/arm_compute/runtime/CL/CLTypes.h index d298ecd614..931740c47f 100644 --- a/arm_compute/runtime/CL/CLTypes.h +++ b/arm_compute/runtime/CL/CLTypes.h @@ -43,12 +43,12 @@ enum class CLGEMMKernelType /** OpenCL GEMM kernel selection parameters. These information are retrieved to select the GEMM kernel on OpenCL */ struct CLGEMMKernelSelectionParams { - unsigned int m{ 0 }; /**< Number of rows for the lhs matrix. Lhs matrix NOT transposed */ - unsigned int n{ 0 }; /**< Number of columns for the rhs matrix. Rhs matrix NOT transposed */ - unsigned int k{ 0 }; /**< Number of rows for the rhs matrix. Rhs matrix NOT transposed */ - unsigned int b{ 0 }; /**< Batch size */ - bool is_rhs_constant{ false }; /**< True if the content of the rhs matrix is constant */ - DataType data_type{ DataType::UNKNOWN }; /**< Data type */ + unsigned int m{0}; /**< Number of rows for the lhs matrix. Lhs matrix NOT transposed */ + unsigned int n{0}; /**< Number of columns for the rhs matrix. Rhs matrix NOT transposed */ + unsigned int k{0}; /**< Number of rows for the rhs matrix. Rhs matrix NOT transposed */ + unsigned int b{0}; /**< Batch size */ + bool is_rhs_constant{false}; /**< True if the content of the rhs matrix is constant */ + DataType data_type{DataType::UNKNOWN}; /**< Data type */ }; /** List the possible OpenCL backends */ diff --git a/arm_compute/runtime/CL/ICLGEMMKernelSelection.h b/arm_compute/runtime/CL/ICLGEMMKernelSelection.h index 7be9393388..5a71a61203 100644 --- a/arm_compute/runtime/CL/ICLGEMMKernelSelection.h +++ b/arm_compute/runtime/CL/ICLGEMMKernelSelection.h @@ -40,8 +40,7 @@ public: * * @param[in] arch GPU target */ - ICLGEMMKernelSelection(GPUTarget arch) - : _target(arch) + ICLGEMMKernelSelection(GPUTarget arch) : _target(arch) { } /** Default Move Constructor. */ @@ -59,7 +58,8 @@ public: virtual CLGEMMKernelType select_kernel(const CLGEMMKernelSelectionParams ¶ms) = 0; protected: - GPUTarget _target; /**< GPU target could be used to call a dedicated heuristic for each GPU IP for a given GPU architecture */ + GPUTarget + _target; /**< GPU target could be used to call a dedicated heuristic for each GPU IP for a given GPU architecture */ }; } // namespace cl_gemm } // namespace arm_compute diff --git a/arm_compute/runtime/CL/ICLOperator.h b/arm_compute/runtime/CL/ICLOperator.h index 38bcaf32f2..c0826e7733 100644 --- a/arm_compute/runtime/CL/ICLOperator.h +++ b/arm_compute/runtime/CL/ICLOperator.h @@ -25,7 +25,6 @@ #define ARM_COMPUTE_ICLOPERATOR_H #include "arm_compute/core/Types.h" - #include "arm_compute/runtime/IOperator.h" #include "arm_compute/runtime/IRuntimeContext.h" #include "arm_compute/runtime/Types.h" @@ -56,8 +55,8 @@ public: ICLOperator &operator=(ICLOperator &&) = default; // Inherited methods overridden: - void run(ITensorPack &tensors) override; - void prepare(ITensorPack &constants) override; + void run(ITensorPack &tensors) override; + void prepare(ITensorPack &constants) override; MemoryRequirements workspace() const override; protected: diff --git a/arm_compute/runtime/CL/functions/CLActivationLayer.h b/arm_compute/runtime/CL/functions/CLActivationLayer.h index 4a718ab4b6..e158efa093 100644 --- a/arm_compute/runtime/CL/functions/CLActivationLayer.h +++ b/arm_compute/runtime/CL/functions/CLActivationLayer.h @@ -24,11 +24,10 @@ #ifndef ARM_COMPUTE_CLACTIVATIONLAYER_H #define ARM_COMPUTE_CLACTIVATIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" #include "arm_compute/runtime/CL/CLRuntimeContext.h" +#include "arm_compute/runtime/IFunction.h" namespace arm_compute { @@ -91,7 +90,10 @@ public: * @param[out] output Destination tensor. Data type supported: same as @p input * @param[in] act_info Activation layer parameters. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, ActivationLayerInfo act_info); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + ActivationLayerInfo act_info); /** Static function to check if given info will lead to a valid configuration of @ref CLActivationLayer * * @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result diff --git a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h index ce5bee8d95..d340d20a1f 100644 --- a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h +++ b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h @@ -91,7 +91,11 @@ public: * @param[out] output Output source tensor. Data types supported: U32/S32. * @param[in] op Reduction operation to perform. Operations supported: ARG_IDX_MAX, ARG_IDX_MIN */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, int axis, ICLTensor *output, const ReductionOperation &op); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + int axis, + ICLTensor *output, + const ReductionOperation &op); /** Static function to check if given info will lead to a valid configuration of @ref CLArgMinMaxLayer * * @param[in] input Input source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/S32/F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h index 37a0680709..f57bc8fe8b 100644 --- a/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_CLBATCHNORMALIZATIONLAYER_H #define ARM_COMPUTE_CLBATCHNORMALIZATIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -84,7 +83,13 @@ public: * @param[in] epsilon (Optional) Small value to avoid division with zero. Default value is 0.001f. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported. */ - void configure(ICLTensor *input, ICLTensor *output, const ICLTensor *mean, const ICLTensor *var, const ICLTensor *beta = nullptr, const ICLTensor *gamma = nullptr, float epsilon = 0.001f, + void configure(ICLTensor *input, + ICLTensor *output, + const ICLTensor *mean, + const ICLTensor *var, + const ICLTensor *beta = nullptr, + const ICLTensor *gamma = nullptr, + float epsilon = 0.001f, ActivationLayerInfo act_info = ActivationLayerInfo()); /** Set the input and output tensors. * @@ -102,9 +107,15 @@ public: * @param[in] epsilon (Optional) Small value to avoid division with zero. Default value is 0.001f. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const ICLTensor *mean, const ICLTensor *var, const ICLTensor *beta = nullptr, - const ICLTensor *gamma = nullptr, - float epsilon = 0.001f, ActivationLayerInfo act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + const ICLTensor *mean, + const ICLTensor *var, + const ICLTensor *beta = nullptr, + const ICLTensor *gamma = nullptr, + float epsilon = 0.001f, + ActivationLayerInfo act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLBatchNormalizationLayer * * @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result. @@ -120,10 +131,14 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const ITensorInfo *mean, const ITensorInfo *var, - const ITensorInfo *beta = nullptr, const ITensorInfo *gamma = nullptr, - float epsilon = 0.001f, ActivationLayerInfo act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const ITensorInfo *mean, + const ITensorInfo *var, + const ITensorInfo *beta = nullptr, + const ITensorInfo *gamma = nullptr, + float epsilon = 0.001f, + ActivationLayerInfo act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h b/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h index 861330b9d4..20b9fdafed 100644 --- a/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h +++ b/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLBATCHTOSPACELAYER_H #define ARM_COMPUTE_CLBATCHTOSPACELAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -82,7 +81,10 @@ public: * @deprecated This method for dynamic block shape is not fully mature and will be removed in 23.08 release */ ARM_COMPUTE_DEPRECATED_REL(23.05) - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *block_shape, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *block_shape, + ICLTensor *output); /** Set the input and output tensors. (Static block shape). * * @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: All. @@ -91,7 +93,11 @@ public: * @param[out] output Tensor output. Data types supported: same as @p input * @param[in] crop_info Information about how the output shape is cropped after batch to space is performed */ - void configure(const ICLTensor *input, int32_t block_shape_x, int32_t block_shape_y, ICLTensor *output, const CropInfo &crop_info = CropInfo{}); + void configure(const ICLTensor *input, + int32_t block_shape_x, + int32_t block_shape_y, + ICLTensor *output, + const CropInfo &crop_info = CropInfo{}); /** Set the input and output tensors. (Static block shape). * * @param[in] compile_context The compile context to be used. @@ -101,7 +107,12 @@ public: * @param[out] output Tensor output. Data types supported: same as @p input * @param[in] crop_info Information about how the output shape is cropped after batch to space is performed */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, int32_t block_shape_x, int32_t block_shape_y, ICLTensor *output, const CropInfo &crop_info = CropInfo{}); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + int32_t block_shape_x, + int32_t block_shape_y, + ICLTensor *output, + const CropInfo &crop_info = CropInfo{}); /** Static function to check if given info will lead to a valid configuration of @ref CLBatchToSpaceLayer * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -124,7 +135,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, int32_t block_shape_x, int32_t block_shape_y, const ITensorInfo *output, const CropInfo &crop_info = CropInfo{}); + static Status validate(const ITensorInfo *input, + int32_t block_shape_x, + int32_t block_shape_y, + const ITensorInfo *output, + const CropInfo &crop_info = CropInfo{}); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLBitwiseAnd.h b/arm_compute/runtime/CL/functions/CLBitwiseAnd.h index b30be9b24f..f82af3af9b 100644 --- a/arm_compute/runtime/CL/functions/CLBitwiseAnd.h +++ b/arm_compute/runtime/CL/functions/CLBitwiseAnd.h @@ -61,7 +61,10 @@ public: * @param[in] input2 Input tensor. Data types supported: U8. * @param[out] output Output tensor. Data types supported: U8. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output); }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLBITWISEAND_H */ diff --git a/arm_compute/runtime/CL/functions/CLBitwiseNot.h b/arm_compute/runtime/CL/functions/CLBitwiseNot.h index 1456ebe57e..31f8e86802 100644 --- a/arm_compute/runtime/CL/functions/CLBitwiseNot.h +++ b/arm_compute/runtime/CL/functions/CLBitwiseNot.h @@ -60,5 +60,5 @@ public: */ void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output); }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLBITWISENOT_H */ diff --git a/arm_compute/runtime/CL/functions/CLBitwiseOr.h b/arm_compute/runtime/CL/functions/CLBitwiseOr.h index ff0a1f0d73..9a25a2099e 100644 --- a/arm_compute/runtime/CL/functions/CLBitwiseOr.h +++ b/arm_compute/runtime/CL/functions/CLBitwiseOr.h @@ -61,7 +61,10 @@ public: * @param[in] input2 Input tensor. Data types supported: U8. * @param[out] output Output tensor. Data types supported: U8. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output); }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLBITWISEOR_H */ diff --git a/arm_compute/runtime/CL/functions/CLBitwiseXor.h b/arm_compute/runtime/CL/functions/CLBitwiseXor.h index 0cd9d073b4..9e288ef7b6 100644 --- a/arm_compute/runtime/CL/functions/CLBitwiseXor.h +++ b/arm_compute/runtime/CL/functions/CLBitwiseXor.h @@ -61,7 +61,10 @@ public: * @param[in] input2 Input tensor. Data types supported: U8. * @param[out] output Output tensor. Data types supported: U8. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output); }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLBITWISEXOR_H */ diff --git a/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h b/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h index d3499c3949..dba5497f5d 100644 --- a/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h +++ b/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h @@ -64,7 +64,10 @@ public: * * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the BoundingBoxTransformInfo struct. */ - void configure(const ICLTensor *boxes, ICLTensor *pred_boxes, const ICLTensor *deltas, const BoundingBoxTransformInfo &info); + void configure(const ICLTensor *boxes, + ICLTensor *pred_boxes, + const ICLTensor *deltas, + const BoundingBoxTransformInfo &info); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -76,7 +79,11 @@ public: * * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the BoundingBoxTransformInfo struct. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *boxes, ICLTensor *pred_boxes, const ICLTensor *deltas, const BoundingBoxTransformInfo &info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *boxes, + ICLTensor *pred_boxes, + const ICLTensor *deltas, + const BoundingBoxTransformInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref CLBoundingBoxTransform * @@ -90,7 +97,10 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *boxes, const ITensorInfo *pred_boxes, const ITensorInfo *deltas, const BoundingBoxTransformInfo &info); + static Status validate(const ITensorInfo *boxes, + const ITensorInfo *pred_boxes, + const ITensorInfo *deltas, + const BoundingBoxTransformInfo &info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLBOUNDINGBOXTRANSFORM_H */ diff --git a/arm_compute/runtime/CL/functions/CLCast.h b/arm_compute/runtime/CL/functions/CLCast.h index 650cd11b9b..9433f08fac 100644 --- a/arm_compute/runtime/CL/functions/CLCast.h +++ b/arm_compute/runtime/CL/functions/CLCast.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLCAST_H #define ARM_COMPUTE_CLCAST_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -79,7 +78,8 @@ public: */ void configure(const ICLTensor *input, ICLTensor *output, ConvertPolicy policy); // Initialize the function's source, destination - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, ConvertPolicy policy); + void + configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, ConvertPolicy policy); /** Static function to check if given info will lead to a valid configuration of @ref CLCast * * @param[in] input Source tensor info. Data types supported: U8/S8/U16/S16/U32/S32/U64/S64/F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h index 3dc62595d2..8ca848a020 100644 --- a/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h +++ b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h @@ -65,7 +65,10 @@ public: * @param[out] output Output tensor. Data type supported: Same as @p input * @param[in] num_groups Number of groups. Must be greater than 1 and the number of channels of the tensors must be a multiple of the number of groups. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, unsigned int num_groups); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + unsigned int num_groups); /** Static function to check if given info will lead to a valid configuration of @ref CLChannelShuffleLayerKernel * * @param[in] input Input tensor info. Data types supported: All. diff --git a/arm_compute/runtime/CL/functions/CLComparison.h b/arm_compute/runtime/CL/functions/CLComparison.h index 3f984900ee..fca4b168b0 100644 --- a/arm_compute/runtime/CL/functions/CLComparison.h +++ b/arm_compute/runtime/CL/functions/CLComparison.h @@ -66,7 +66,11 @@ public: * @param[out] output Destination tensor. Data types supported: U8. * @param[out] operation Comparison operation to be used. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, ComparisonOperation operation); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + ComparisonOperation operation); /** Static function to check if given info will lead to a valid configuration of @ref CLComparison * * @param[in] input1 Source tensor. Data types supported: All. @@ -76,7 +80,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ComparisonOperation operation); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + ComparisonOperation operation); }; /** Basic function to run @ref CLComparisonKernel */ diff --git a/arm_compute/runtime/CL/functions/CLConcatenateLayer.h b/arm_compute/runtime/CL/functions/CLConcatenateLayer.h index 71e84e21b5..88c4bed595 100644 --- a/arm_compute/runtime/CL/functions/CLConcatenateLayer.h +++ b/arm_compute/runtime/CL/functions/CLConcatenateLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLCONCATENATELAYER_H #define ARM_COMPUTE_CLCONCATENATELAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> #include <vector> @@ -95,7 +94,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input. * @param[in] axis Concatenation axis. Supported underlying concatenation axis are 0, 1, 2 and 3. */ - void configure(const CLCompileContext &compile_context, std::vector<const ICLTensor *> &inputs_vector, ICLTensor *output, size_t axis); + void configure(const CLCompileContext &compile_context, + std::vector<const ICLTensor *> &inputs_vector, + ICLTensor *output, + size_t axis); /** Static function to check if given info will lead to a valid configuration of @ref CLConcatenateLayer * * @note Input and output tensor dimensions preconditions defer depending on the concatenation axis. @@ -108,7 +110,8 @@ public: * * @return a status */ - static Status validate(const std::vector<const ITensorInfo *> &inputs_vector, const ITensorInfo *output, size_t axis); + static Status + validate(const std::vector<const ITensorInfo *> &inputs_vector, const ITensorInfo *output, size_t axis); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLConv3D.h b/arm_compute/runtime/CL/functions/CLConv3D.h index 5728fe79d8..aabaf01ab7 100644 --- a/arm_compute/runtime/CL/functions/CLConv3D.h +++ b/arm_compute/runtime/CL/functions/CLConv3D.h @@ -77,20 +77,33 @@ public: * @param[in] conv3d_info Contains strides, padding, rounding, activation, dilation and fast math information. Activation and fast math are currently unused. * */ - void configure(const CLCompileContext &compile_context, const ICLTensor *src, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *dst, const Conv3dInfo &conv3d_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *src, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *dst, + const Conv3dInfo &conv3d_info); /** Set the src and dst tensors. * * Similar to CLConv3D::configure() but using the default compile context * */ - void configure(const ICLTensor *src, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *dst, const Conv3dInfo &conv3d_info); + void configure(const ICLTensor *src, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *dst, + const Conv3dInfo &conv3d_info); /** Static function to check if given info will lead to a valid configuration of @ref CLConv3D * * Similar to CLConv3D::configure() * * @return a status */ - static Status validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, const Conv3dInfo &conv3d_info); + static Status validate(const ITensorInfo *src, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *dst, + const Conv3dInfo &conv3d_info); // Inherited methods overridden: void run() override; @@ -99,5 +112,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLCONVOLUTION3DLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h index 0a634b5482..409430d595 100644 --- a/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h +++ b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h @@ -69,7 +69,10 @@ public: * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). * @param[in] data_layout The data layout the weights have been trained in. */ - void configure(const ICLTensor *input, ICLTensor *output, const TensorShape &original_input_shape, DataLayout data_layout); + void configure(const ICLTensor *input, + ICLTensor *output, + const TensorShape &original_input_shape, + DataLayout data_layout); /** Initialize the function. * * @param[in] compile_context The compile context to be used. @@ -78,7 +81,11 @@ public: * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). * @param[in] data_layout The data layout the weights have been trained in. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const TensorShape &original_input_shape, DataLayout data_layout); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const TensorShape &original_input_shape, + DataLayout data_layout); /** Static function to check if given info will lead to a valid configuration of @ref CLConvertFullyConnectedWeights * * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: All. @@ -86,7 +93,10 @@ public: * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). * @param[in] data_layout The data layout the weights have been trained in. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const TensorShape &original_input_shape, DataLayout data_layout); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const TensorShape &original_input_shape, + DataLayout data_layout); // Inherited methods overridden: void run() override; @@ -144,7 +154,10 @@ public: * @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer). * @param[in] data_layout The data layout the weights have been trained in. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const TensorShape &original_input_shape, DataLayout data_layout) + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const TensorShape &original_input_shape, + DataLayout data_layout) { _func.configure(compile_context, input, &_output, original_input_shape, data_layout); } diff --git a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h index 77bf48d613..8487be71c3 100644 --- a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h @@ -120,8 +120,16 @@ public: * available which may introduce a drop of accuracy as well. Default is false * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is only supported for NCHW data layout */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, const WeightsInfo &weights_info = WeightsInfo(), - const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, unsigned int num_groups = 1); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -142,9 +150,17 @@ public: * available which may introduce a drop of accuracy as well. Default is false * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is only supported for NCHW data layout */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, - unsigned int num_groups = 1); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if given info will lead to a valid configuration of @ref CLConvolutionLayer * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], @@ -166,9 +182,16 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false, - unsigned int num_groups = 1); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false, + unsigned int num_groups = 1); /** Static function to check if given info will return the convolution called by @ref CLConvolutionLayer * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], @@ -188,8 +211,15 @@ public: * * @return the Convolution Method Hint */ - static ConvolutionMethod get_convolution_method(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info, const ActivationLayerInfo &act_info, const GPUTarget gpu_target, const Size2D &dilation = Size2D(1U, 1U), bool enable_fast_math = false); + static ConvolutionMethod get_convolution_method(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info, + const ActivationLayerInfo &act_info, + const GPUTarget gpu_target, + const Size2D &dilation = Size2D(1U, 1U), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; void prepare() override; diff --git a/arm_compute/runtime/CL/functions/CLCopy.h b/arm_compute/runtime/CL/functions/CLCopy.h index 4fc4183d3e..fd40b7b9de 100644 --- a/arm_compute/runtime/CL/functions/CLCopy.h +++ b/arm_compute/runtime/CL/functions/CLCopy.h @@ -27,6 +27,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/Window.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -74,7 +75,10 @@ public: * @param[in] dst_window (Optional) Window to be used in case only copying into part of a tensor. Default is nullptr. * */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, Window *dst_window = nullptr); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + Window *dst_window = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLCopy * * @param[in] input Source tensor. Data types supported: All. diff --git a/arm_compute/runtime/CL/functions/CLCrop.h b/arm_compute/runtime/CL/functions/CLCrop.h index a474215190..2942e9362a 100644 --- a/arm_compute/runtime/CL/functions/CLCrop.h +++ b/arm_compute/runtime/CL/functions/CLCrop.h @@ -27,6 +27,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/Window.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -71,7 +72,13 @@ public: * @param[in] extrapolation_value Value to be used for values outside of the image. Default is 0. * @param[in] output_window Output window to be used in case cropped image is being copied into a tensor. Default is nullptr. */ - void configure(const ICLTensor *input, ICLTensor *output, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value = 0, Window *output_window = nullptr); + void configure(const ICLTensor *input, + ICLTensor *output, + Coordinates2D start, + Coordinates2D end, + uint32_t batch_index, + float extrapolation_value = 0, + Window *output_window = nullptr); /** Configure function * * @note Supported tensor rank: up to 4 @@ -85,8 +92,14 @@ public: * @param[in] extrapolation_value Value to be used for values outside of the image. Default is 0. * @param[in] output_window Output window to be used in case cropped image is being copied into a tensor. Default is nullptr. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value = 0, - Window *output_window = nullptr); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + Coordinates2D start, + Coordinates2D end, + uint32_t batch_index, + float extrapolation_value = 0, + Window *output_window = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLStridedSliceKernel * @@ -100,8 +113,13 @@ public: * @param[in] extrapolation_value Value to be used for values outside of the image. Default is 0. * @param[in] output_window Output window to be used in case cropped image is being copied into a tensor. Default is nullptr. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, Coordinates2D start, Coordinates2D end, uint32_t batch_index, float extrapolation_value = 0, - Window *output_window = nullptr); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + Coordinates2D start, + Coordinates2D end, + uint32_t batch_index, + float extrapolation_value = 0, + Window *output_window = nullptr); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLCropResize.h b/arm_compute/runtime/CL/functions/CLCropResize.h index 5c60c2879c..6fb055e893 100644 --- a/arm_compute/runtime/CL/functions/CLCropResize.h +++ b/arm_compute/runtime/CL/functions/CLCropResize.h @@ -25,7 +25,6 @@ #define ARM_COMPUTE_CL_CROP_RESIZE_H #include "arm_compute/core/CL/ICLTensor.h" - #include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/CL/functions/CLCopy.h" #include "arm_compute/runtime/CL/functions/CLCrop.h" @@ -82,8 +81,13 @@ public: * @param[in] method The policy to be used when resizing image. Default is bilinear. * @param[in] extrapolation_value Value to be used for values outside of the image for cropping and resizing. Default is 0. */ - void configure(const ICLTensor *input, ICLTensor *boxes, ICLTensor *box_ind, ICLTensor *output, Coordinates2D crop_size, - InterpolationPolicy method = InterpolationPolicy::BILINEAR, float extrapolation_value = 0); + void configure(const ICLTensor *input, + ICLTensor *boxes, + ICLTensor *box_ind, + ICLTensor *output, + Coordinates2D crop_size, + InterpolationPolicy method = InterpolationPolicy::BILINEAR, + float extrapolation_value = 0); /** Configure kernel * * @note Supported tensor rank: up to 4 @@ -100,8 +104,14 @@ public: * @param[in] method The policy to be used when resizing image. Default is bilinear. * @param[in] extrapolation_value Value to be used for values outside of the image for cropping and resizing. Default is 0. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *boxes, ICLTensor *box_ind, ICLTensor *output, Coordinates2D crop_size, - InterpolationPolicy method = InterpolationPolicy::BILINEAR, float extrapolation_value = 0); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *boxes, + ICLTensor *box_ind, + ICLTensor *output, + Coordinates2D crop_size, + InterpolationPolicy method = InterpolationPolicy::BILINEAR, + float extrapolation_value = 0); /** Static function to check if given info will lead to a valid configuration of @ref NESlice * @@ -121,8 +131,13 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, ITensorInfo *boxes, ITensorInfo *box_ind, const ITensorInfo *output, - Coordinates2D crop_size, InterpolationPolicy method, float extrapolation_value); + static Status validate(const ITensorInfo *input, + ITensorInfo *boxes, + ITensorInfo *box_ind, + const ITensorInfo *output, + Coordinates2D crop_size, + InterpolationPolicy method, + float extrapolation_value); void run() override; diff --git a/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h index 0c59e2c86d..92f87ee461 100644 --- a/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h @@ -70,7 +70,12 @@ public: * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. * */ - void configure(ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info, const WeightsInfo &weights_info = WeightsInfo()); + void configure(ICLTensor *input, + ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info, + const WeightsInfo &weights_info = WeightsInfo()); /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. @@ -82,8 +87,13 @@ public: * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. * */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info, - const WeightsInfo &weights_info = WeightsInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info, + const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLDeconvolutionLayer * * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32. @@ -95,11 +105,19 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, ITensorInfo *output, const PadStrideInfo &deconv_info, - const WeightsInfo &weights_info = WeightsInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + ITensorInfo *output, + const PadStrideInfo &deconv_info, + const WeightsInfo &weights_info = WeightsInfo()); - static DeconvolutionMethod get_deconvolution_method(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, ITensorInfo *output, const PadStrideInfo &deconv_info, - const WeightsInfo &weights_info); + static DeconvolutionMethod get_deconvolution_method(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + ITensorInfo *output, + const PadStrideInfo &deconv_info, + const WeightsInfo &weights_info); // Inherited methods overridden: void run() override; void prepare() override; diff --git a/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h b/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h index 344ebd0afb..5a2abafe79 100644 --- a/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h +++ b/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h @@ -82,7 +82,8 @@ public: * @param[out] output Destination tensor. Data type supported: same as @p input. * @param[in] info Contains padding and policies to be used in the deconvolution. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const PadStrideInfo &info); + void + configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const PadStrideInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref CLDeconvolutionLayerUpsample * * @param[in] input Source tensor info. Data type supported: All. diff --git a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h index 58deb7ec40..3e7ca8830b 100644 --- a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLDEPTHCONVERT_H #define ARM_COMPUTE_CLDEPTHCONVERT_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -96,7 +95,11 @@ public: * @param[in] policy Conversion policy. * @param[in] shift Value for down/up conversions. Must be 0 <= shift < 8. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, ConvertPolicy policy, uint32_t shift); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + ConvertPolicy policy, + uint32_t shift); /** Static function to check if given info will lead to a valid configuration of @ref CLDepthConvertLayer * * @param[in] input Source tensor info. Data types supported: U8/S8/U16/S16/U32/S32/F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h index 0026cc2b67..14d0a7ec7c 100644 --- a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h @@ -60,7 +60,8 @@ public: * @param[out] output Tensor output. Data types supported: same as @p input * @param[in] block_shape Block shape value. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, int32_t block_shape); + void + configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, int32_t block_shape); /** Static function to check if given info will lead to a valid configuration of @ref CLDepthToSpaceLayer. * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -71,5 +72,5 @@ public: */ static Status validate(const ITensorInfo *input, const ITensorInfo *output, int32_t block_shape); }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLDEPTHTOSPACELAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h index 2798449100..2c0fa7aa22 100644 --- a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h @@ -88,15 +88,28 @@ public: * * @note: For in-place support, please check @ref CLDepthwiseConvolutionLayerNativeKernel */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, ActivationLayerInfo act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + ActivationLayerInfo act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); /** Initialize the function's source, destination, weights and convolution information. * * Similar to @ref CLDepthwiseConvolutionLayer::configure() */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, ActivationLayerInfo act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + ActivationLayerInfo act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); /** Static function to check if given info will lead to a valid configuration of @ref CLDepthwiseConvolutionLayer * @@ -104,8 +117,14 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - unsigned int depth_multiplier = 1, ActivationLayerInfo act_info = ActivationLayerInfo(), const Size2D &dilation = Size2D(1U, 1U)); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + unsigned int depth_multiplier = 1, + ActivationLayerInfo act_info = ActivationLayerInfo(), + const Size2D &dilation = Size2D(1U, 1U)); // Inherited methods overriden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h index 462a3ac07e..84900b03a3 100644 --- a/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDirectConvolutionLayer.h @@ -79,7 +79,12 @@ public: * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -94,7 +99,12 @@ public: * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLDirectConvolutionLayer * @@ -111,7 +121,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: @@ -121,5 +135,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLDIRECTCONVOLUTIONLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLDirectDeconvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDirectDeconvolutionLayer.h index d0a61cdd36..14384a09b5 100644 --- a/arm_compute/runtime/CL/functions/CLDirectDeconvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLDirectDeconvolutionLayer.h @@ -24,12 +24,11 @@ #ifndef ARM_COMPUTE_CLDIRECTDECONVOLUTIONLAYER_H #define ARM_COMPUTE_CLDIRECTDECONVOLUTIONLAYER_H +#include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/CL/functions/CLConvolutionLayer.h" #include "arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h" #include "arm_compute/runtime/CL/functions/CLReverse.h" #include "arm_compute/runtime/CL/functions/CLTranspose.h" - -#include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" @@ -111,7 +110,12 @@ public: * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. * */ - void configure(ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &info, const WeightsInfo &weights_info = WeightsInfo()); + void configure(ICLTensor *input, + ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &info, + const WeightsInfo &weights_info = WeightsInfo()); /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. @@ -125,8 +129,13 @@ public: * @param[in] weights_info (Optional) Weights information needed for @ref CLConvolutionLayer, specifies if the weights tensor has been reshaped with @ref opencl::kernels::ClWeightsReshapeKernel. * */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &info, - const WeightsInfo &weights_info = WeightsInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &info, + const WeightsInfo &weights_info = WeightsInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLDirectDeconvolutionLayer * * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. @@ -140,8 +149,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, ITensorInfo *output, const PadStrideInfo &info, - const WeightsInfo &weights_info = WeightsInfo()); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + ITensorInfo *output, + const PadStrideInfo &info, + const WeightsInfo &weights_info = WeightsInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLElementwiseOperations.h b/arm_compute/runtime/CL/functions/CLElementwiseOperations.h index 9de362d2b2..13844c98a1 100644 --- a/arm_compute/runtime/CL/functions/CLElementwiseOperations.h +++ b/arm_compute/runtime/CL/functions/CLElementwiseOperations.h @@ -82,7 +82,11 @@ public: * @param[in] policy Policy to use to handle overflow. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * Valid configurations (Input1,Input2) -> Output : @@ -108,7 +112,11 @@ public: * @param[in] policy Policy to use to handle overflow. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy, + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output, + ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClSaturatedArithmeticKernel for addition * @@ -134,7 +142,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -192,7 +204,11 @@ public: * @param[in] policy Policy to use to handle overflow. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * Valid configurations (Input1,Input2) -> Output : @@ -218,7 +234,11 @@ public: * @param[in] policy Policy to use to handle overflow. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy, + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output, + ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClSaturatedArithmeticKernel for subtraction * @@ -244,7 +264,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -292,7 +316,10 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output. * * @param[in] compile_context The compile context to be used. @@ -303,7 +330,11 @@ public: * @param[out] output Output tensor. Data types supported: Same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticDivision * * @param[in] input1 First tensor input info. Data types supported: F16/F32. @@ -313,7 +344,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -368,7 +402,10 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * @param[in] compile_context The compile context to be used. @@ -379,7 +416,11 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClArithmeticKernel for max * * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/S32/U32/F16/F32. @@ -389,7 +430,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -444,7 +488,10 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * @param[in] compile_context The compile context to be used. @@ -455,7 +502,11 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClArithmeticKernel for min * * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/S32/U32/F16/F32. @@ -465,7 +516,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -518,7 +572,10 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * @param[in] compile_context The compile context to be used. @@ -529,7 +586,11 @@ public: * @param[out] output Output tensor. Data types supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClArithmeticKernel for squared difference * * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/F16/F32. @@ -539,7 +600,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -587,7 +651,10 @@ public: * @param[out] output Output tensor. Data types supported:F16/F32. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and conversion policy. * * @param[in] compile_context The compile context to be used. @@ -598,7 +665,11 @@ public: * @param[out] output Output tensor. Data types supported:F16/F32. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClArithmeticKernel for power * * @param[in] input1 First tensor input info. Data types supported: F16/F32. @@ -608,7 +679,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLElementwiseUnaryLayer.h b/arm_compute/runtime/CL/functions/CLElementwiseUnaryLayer.h index 594ee4cfdc..d186b70d93 100644 --- a/arm_compute/runtime/CL/functions/CLElementwiseUnaryLayer.h +++ b/arm_compute/runtime/CL/functions/CLElementwiseUnaryLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H #define ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/CL/functions/CLFFT1D.h b/arm_compute/runtime/CL/functions/CLFFT1D.h index c7112dc737..49ecf3c260 100644 --- a/arm_compute/runtime/CL/functions/CLFFT1D.h +++ b/arm_compute/runtime/CL/functions/CLFFT1D.h @@ -24,10 +24,9 @@ #ifndef ARM_COMPUTE_CLFFT1D_H #define ARM_COMPUTE_CLFFT1D_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/FunctionDescriptors.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" namespace arm_compute @@ -82,7 +81,10 @@ public: * @param[out] output Destination tensor. Data types and data layouts supported: Same as @p input. * @param[in] config FFT related configuration */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const FFT1DInfo &config); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const FFT1DInfo &config); /** Static function to check if given info will lead to a valid configuration of @ref CLFFT1D. * * @param[in] input Source tensor info. Data types supported: F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLFFT2D.h b/arm_compute/runtime/CL/functions/CLFFT2D.h index 3d20327bf1..b7d15f1602 100644 --- a/arm_compute/runtime/CL/functions/CLFFT2D.h +++ b/arm_compute/runtime/CL/functions/CLFFT2D.h @@ -24,11 +24,10 @@ #ifndef ARM_COMPUTE_CLFFT2D_H #define ARM_COMPUTE_CLFFT2D_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/CL/functions/CLFFT1D.h" #include "arm_compute/runtime/FunctionDescriptors.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/MemoryGroup.h" namespace arm_compute @@ -79,7 +78,10 @@ public: * @param[out] output Destination tensor. Data types and data layouts supported: Same as @p input. * @param[in] config FFT related configuration */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const FFT2DInfo &config); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const FFT2DInfo &config); /** Static function to check if given info will lead to a valid configuration of @ref CLFFT2D. * * @param[in] input Source tensor info. Data types supported: F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLFFTConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLFFTConvolutionLayer.h index f873cb0b86..ed78bbb7a7 100644 --- a/arm_compute/runtime/CL/functions/CLFFTConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLFFTConvolutionLayer.h @@ -24,8 +24,6 @@ #ifndef ARM_COMPUTE_CLFFTCONVOLUTIONLAYER_H #define ARM_COMPUTE_CLFFTCONVOLUTIONLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/runtime/CL/functions/CLActivationLayer.h" #include "arm_compute/runtime/CL/functions/CLElementwiseOperations.h" @@ -37,6 +35,7 @@ #include "arm_compute/runtime/CL/functions/CLReshapeLayer.h" #include "arm_compute/runtime/CL/functions/CLReverse.h" #include "arm_compute/runtime/CL/functions/CLSlice.h" +#include "arm_compute/runtime/IFunction.h" namespace arm_compute { @@ -94,8 +93,13 @@ public: * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation * available which may introduce a drop of accuracy as well. Default is false */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); /** Set the input and output tensors. * * @note: This function only works with any square kernel size and unit strides for both NCHW and NHWC data layout @@ -113,8 +117,14 @@ public: * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation * available which may introduce a drop of accuracy as well. Default is false */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); /** Static function to check if given info will lead to a valid configuration of @ref CLFFTConvolutionLayer * * @note: This function only works with any square kernel size and unit strides for both NCHW and NHWC data layout @@ -133,8 +143,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLFill.h b/arm_compute/runtime/CL/functions/CLFill.h index 341d93a9f6..be1059761a 100644 --- a/arm_compute/runtime/CL/functions/CLFill.h +++ b/arm_compute/runtime/CL/functions/CLFill.h @@ -28,6 +28,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/core/Window.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -73,7 +74,10 @@ public: * @param[in] constant_value The value used to fill the planes of the tensor * @param[in] window Window to be used in case setting only part of a tensor. Default is nullptr. */ - void configure(const CLCompileContext &compile_context, ICLTensor *tensor, const PixelValue &constant_value, Window *window = nullptr); + void configure(const CLCompileContext &compile_context, + ICLTensor *tensor, + const PixelValue &constant_value, + Window *window = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLFill * * @param[in] tensor Source tensor info. Data types supported: All. diff --git a/arm_compute/runtime/CL/functions/CLFloor.h b/arm_compute/runtime/CL/functions/CLFloor.h index 87cd5b44c7..4d3d704857 100644 --- a/arm_compute/runtime/CL/functions/CLFloor.h +++ b/arm_compute/runtime/CL/functions/CLFloor.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLFLOOR_H #define ARM_COMPUTE_CLFLOOR_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> diff --git a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h index b784226a2f..9fd0b4aaef 100644 --- a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h +++ b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h @@ -25,9 +25,8 @@ #define ARM_COMPUTE_CLFULLYCONNECTEDLAYER_H #include "arm_compute/function_info/FullyConnectedLayerInfo.h" -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/CL/CLTensor.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IWeightsManager.h" #include "arm_compute/runtime/MemoryGroup.h" @@ -45,7 +44,8 @@ class CLFullyConnectedLayer : public IFunction { public: /** Constructor */ - CLFullyConnectedLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, IWeightsManager *weights_manager = nullptr); + CLFullyConnectedLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, + IWeightsManager *weights_manager = nullptr); /** Default destructor */ ~CLFullyConnectedLayer(); /** Prevent instances of this class from being copied (As this class contains pointers) */ @@ -83,13 +83,20 @@ public: * Data type supported: Same as @p input. * @param[in] fc_info (Optional) Fully connected layer additional info */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo()); /** Set the input and output tensors. * * Similar to @ref CLFullyConnectedLayer */ - void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, + void configure(const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLFullyConnectedLayer * @@ -97,7 +104,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, FullyConnectedLayerInfo fc_info = FullyConnectedLayerInfo()); //Inherited methods override diff --git a/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h b/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h index cd75270392..2e777273cd 100644 --- a/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h +++ b/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h @@ -78,9 +78,16 @@ public: * @param[in] epsilon (Optional) Batch normalization layer epsilon parameter. Defaults to 0.001f. * @param[in] fbn_type (Optional) Fused batch normalization type. Defaults to Convolution. */ - void configure(const ICLTensor *input_weights, const ICLTensor *bn_mean, const ICLTensor *bn_var, ICLTensor *fused_weights, ICLTensor *fused_bias, - const ICLTensor *input_bias = nullptr, const ICLTensor *bn_beta = nullptr, const ICLTensor *bn_gamma = nullptr, - float epsilon = 0.001f, FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); + void configure(const ICLTensor *input_weights, + const ICLTensor *bn_mean, + const ICLTensor *bn_var, + ICLTensor *fused_weights, + ICLTensor *fused_bias, + const ICLTensor *input_bias = nullptr, + const ICLTensor *bn_beta = nullptr, + const ICLTensor *bn_gamma = nullptr, + float epsilon = 0.001f, + FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -97,9 +104,17 @@ public: * @param[in] epsilon (Optional) Batch normalization layer epsilon parameter. Defaults to 0.001f. * @param[in] fbn_type (Optional) Fused batch normalization type. Defaults to Convolution. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input_weights, const ICLTensor *bn_mean, const ICLTensor *bn_var, ICLTensor *fused_weights, ICLTensor *fused_bias, - const ICLTensor *input_bias = nullptr, const ICLTensor *bn_beta = nullptr, const ICLTensor *bn_gamma = nullptr, - float epsilon = 0.001f, FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input_weights, + const ICLTensor *bn_mean, + const ICLTensor *bn_var, + ICLTensor *fused_weights, + ICLTensor *fused_bias, + const ICLTensor *input_bias = nullptr, + const ICLTensor *bn_beta = nullptr, + const ICLTensor *bn_gamma = nullptr, + float epsilon = 0.001f, + FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); /** Static function to check if given info will lead to a valid configuration of @ref CLFuseBatchNormalization * * @param[in] input_weights Input weights tensor info for convolution or depthwise convolution layer. Data type supported: F16/F32. Data layout supported: NCHW, NHWC @@ -117,10 +132,16 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input_weights, const ITensorInfo *bn_mean, const ITensorInfo *bn_var, - const ITensorInfo *fused_weights, const ITensorInfo *fused_bias, - const ITensorInfo *input_bias = nullptr, const ITensorInfo *bn_beta = nullptr, const ITensorInfo *bn_gamma = nullptr, - float epsilon = 0.001f, FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); + static Status validate(const ITensorInfo *input_weights, + const ITensorInfo *bn_mean, + const ITensorInfo *bn_var, + const ITensorInfo *fused_weights, + const ITensorInfo *fused_bias, + const ITensorInfo *input_bias = nullptr, + const ITensorInfo *bn_beta = nullptr, + const ITensorInfo *bn_gamma = nullptr, + float epsilon = 0.001f, + FuseBatchNormalizationType fbn_type = FuseBatchNormalizationType::CONVOLUTION); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGEMM.h b/arm_compute/runtime/CL/functions/CLGEMM.h index 3a39aca692..f5e6aa1237 100644 --- a/arm_compute/runtime/CL/functions/CLGEMM.h +++ b/arm_compute/runtime/CL/functions/CLGEMM.h @@ -92,13 +92,26 @@ public: * if the reshape of matrix B should happen only for the first run. GEMMInfo also contains information about the reshaping * in case matrix A and matrix B have been already transformed. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *a, const ICLTensor *b, const ICLTensor *c, ICLTensor *output, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + void configure(const CLCompileContext &compile_context, + const ICLTensor *a, + const ICLTensor *b, + const ICLTensor *c, + ICLTensor *output, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); /** Initialise the kernel's inputs and output * * Similar to @ref CLGEMM::configure() */ - void configure(const ICLTensor *a, const ICLTensor *b, const ICLTensor *c, ICLTensor *output, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + void configure(const ICLTensor *a, + const ICLTensor *b, + const ICLTensor *c, + ICLTensor *output, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMM. * @@ -106,7 +119,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, float alpha, float beta, const GEMMInfo &gemm_info = GEMMInfo()); + static Status validate(const ITensorInfo *a, + const ITensorInfo *b, + const ITensorInfo *c, + const ITensorInfo *output, + float alpha, + float beta, + const GEMMInfo &gemm_info = GEMMInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h index 4bafef27a9..70ceb1513b 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h @@ -52,7 +52,8 @@ public: * @param[in] memory_manager (Optional) Memory manager. * @param[in] weights_manager (Optional) Weights manager. */ - CLGEMMConvolutionLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, IWeightsManager *weights_manager = nullptr); + CLGEMMConvolutionLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr, + IWeightsManager *weights_manager = nullptr); /** Prevent instances of this class from being copied (As this class contains pointers) */ CLGEMMConvolutionLayer(const CLGEMMConvolutionLayer &) = delete; /** Default move constructor */ @@ -95,8 +96,15 @@ public: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is only supported for NCHW data layout */ - void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, const WeightsInfo &weights_info = WeightsInfo(), - const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), unsigned int num_groups = 1); + void configure(const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + unsigned int num_groups = 1); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -116,9 +124,16 @@ public: * @param[in] act_info (Optional) Activation layer information in case of a fused activation. * @param[in] num_groups (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is only supported for NCHW data layout */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), - const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), unsigned int num_groups = 1); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + unsigned int num_groups = 1); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMConvolutionLayer. * * @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM], @@ -139,8 +154,15 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const WeightsInfo &weights_info = WeightsInfo(), const Size2D &dilation = Size2D(1U, 1U), const ActivationLayerInfo &act_info = ActivationLayerInfo(), unsigned int num_groups = 1); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const WeightsInfo &weights_info = WeightsInfo(), + const Size2D &dilation = Size2D(1U, 1U), + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + unsigned int num_groups = 1); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h b/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h index c985738a9c..3e8929c5ad 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLGEMMDeconvolutionLayer.h @@ -113,7 +113,11 @@ public: * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. Data layout supported: same as @p input. * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this is described in @ref PadStrideInfo. This function supports only stride_x = weights.width && stride_y = weights.height. Moreover, padding is not supported. */ - void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info); + void configure(const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info); /** Set the input, weights, biases and output tensors. * * @param[in] compile_context The compile context to be used. @@ -124,7 +128,12 @@ public: * @param[out] output Output tensor. The output has the same number of dimensions as the @p input. Data layout supported: same as @p input. * @param[in] deconv_info Contains padding and policies to be used in the deconvolution, this is described in @ref PadStrideInfo. This function supports only stride_x = weights.width && stride_y = weights.height. Moreover, padding is not supported. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *weights, const ICLTensor *bias, ICLTensor *output, const PadStrideInfo &deconv_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *bias, + ICLTensor *output, + const PadStrideInfo &deconv_info); /** Static function to check if given info will lead to a valid configuration of @ref CLDeconvolutionLayer * * @param[in] input Input tensor info. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. @@ -136,7 +145,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *bias, const ITensorInfo *output, const PadStrideInfo &deconv_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *bias, + const ITensorInfo *output, + const PadStrideInfo &deconv_info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h index 8b8d9f235f..1b8e5dcc1d 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h +++ b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h @@ -91,7 +91,11 @@ public: * @param[in] gemm_info (Optional) Specifies if the matrix A and/or matrix B have been reshaped and * if the reshape of matrix B should be executed only for the first run */ - void configure(const ICLTensor *a, const ICLTensor *b, const ICLTensor *c, ICLTensor *output, const GEMMInfo &gemm_info = GEMMInfo()); + void configure(const ICLTensor *a, + const ICLTensor *b, + const ICLTensor *c, + ICLTensor *output, + const GEMMInfo &gemm_info = GEMMInfo()); /** Initialise the kernel's inputs, output * * @note GEMMLowp: low precision GEMM kernel. [A * B + C] @@ -110,7 +114,12 @@ public: * @param[in] gemm_info (Optional) Specifies if the matrix A and/or matrix B have been reshaped and * if the reshape of matrix B should be executed only for the first run */ - void configure(const CLCompileContext &compile_context, const ICLTensor *a, const ICLTensor *b, const ICLTensor *c, ICLTensor *output, const GEMMInfo &gemm_info = GEMMInfo()); + void configure(const CLCompileContext &compile_context, + const ICLTensor *a, + const ICLTensor *b, + const ICLTensor *c, + ICLTensor *output, + const GEMMInfo &gemm_info = GEMMInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLGEMMLowpMatrixMultiplyCore * * @param[in] a First input tensor info (Matrix A). Data type supported: QASYMM8. @@ -122,7 +131,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, const GEMMInfo &gemm_info = GEMMInfo()); + static Status validate(const ITensorInfo *a, + const ITensorInfo *b, + const ITensorInfo *c, + const ITensorInfo *output, + const GEMMInfo &gemm_info = GEMMInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h b/arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h index 6ec7b71f7d..ff9c872896 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h +++ b/arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h @@ -85,7 +85,8 @@ public: * @param[out] output Output tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/QSYMM16 * @param[in] info GEMMLowp output stage metadata. */ - void configure(const ICLTensor *input, const ICLTensor *bias, ICLTensor *output, const GEMMLowpOutputStageInfo &info); + void + configure(const ICLTensor *input, const ICLTensor *bias, ICLTensor *output, const GEMMLowpOutputStageInfo &info); /** Initialise the kernel's inputs, output * * @param[in] compile_context The compile context to be used. @@ -95,7 +96,11 @@ public: * @param[out] output Output tensor. Data type supported: QASYMM8/QASYMM8_SIGNED * @param[in] info GEMMLowp output stage metadata. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *bias, ICLTensor *output, const GEMMLowpOutputStageInfo &info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *bias, + ICLTensor *output, + const GEMMLowpOutputStageInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref opencl::kernels::ClGemmLowpQuantizeDownInt32ScaleByFixedPointKernel * * @param[in] input Input tensor. It is the output of @ref CLGEMMLowpMatrixMultiplyCore function. Data type supported: S32 @@ -106,7 +111,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output, const GEMMLowpOutputStageInfo &info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *bias, + const ITensorInfo *output, + const GEMMLowpOutputStageInfo &info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLGather.h b/arm_compute/runtime/CL/functions/CLGather.h index 0f1ccbad08..360c8757b6 100644 --- a/arm_compute/runtime/CL/functions/CLGather.h +++ b/arm_compute/runtime/CL/functions/CLGather.h @@ -62,7 +62,11 @@ public: * @param[out] output Destination tensor. Data type supported: Same as @p input * @param[in] axis (Optional) The axis in @p input to gather @p indices from. Defaults to 0 */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *indices, ICLTensor *output, int axis = 0); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *indices, + ICLTensor *output, + int axis = 0); /** Static function to check if given info will lead to a valid configuration of @ref CLGatherKernel * @@ -73,7 +77,8 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, int axis = 0); + static Status + validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, int axis = 0); }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLGATHER_H */ diff --git a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h index aec5cdf1a8..3a201e79b0 100644 --- a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h +++ b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h @@ -100,7 +100,12 @@ public: * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the @ref GenerateProposalsInfo struct. * @note Proposals contains all the proposals. Of those, only the first num_valid_proposals are valid. */ - void configure(const ICLTensor *scores, const ICLTensor *deltas, const ICLTensor *anchors, ICLTensor *proposals, ICLTensor *scores_out, ICLTensor *num_valid_proposals, + void configure(const ICLTensor *scores, + const ICLTensor *deltas, + const ICLTensor *anchors, + ICLTensor *proposals, + ICLTensor *scores_out, + ICLTensor *num_valid_proposals, const GenerateProposalsInfo &info); /** Set the input and output tensors. * @@ -118,8 +123,14 @@ public: * @note Only single image prediction is supported. Height and Width (and scale) of the image will be contained in the @ref GenerateProposalsInfo struct. * @note Proposals contains all the proposals. Of those, only the first num_valid_proposals are valid. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *scores, const ICLTensor *deltas, const ICLTensor *anchors, ICLTensor *proposals, ICLTensor *scores_out, - ICLTensor *num_valid_proposals, const GenerateProposalsInfo &info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *scores, + const ICLTensor *deltas, + const ICLTensor *anchors, + ICLTensor *proposals, + ICLTensor *scores_out, + ICLTensor *num_valid_proposals, + const GenerateProposalsInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref CLGenerateProposalsLayer * @@ -135,7 +146,11 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *scores, const ITensorInfo *deltas, const ITensorInfo *anchors, const ITensorInfo *proposals, const ITensorInfo *scores_out, + static Status validate(const ITensorInfo *scores, + const ITensorInfo *deltas, + const ITensorInfo *anchors, + const ITensorInfo *proposals, + const ITensorInfo *scores_out, const ITensorInfo *num_valid_proposals, const GenerateProposalsInfo &info); diff --git a/arm_compute/runtime/CL/functions/CLIndirectConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLIndirectConvolutionLayer.h index 12b83ea25b..91952af5dc 100644 --- a/arm_compute/runtime/CL/functions/CLIndirectConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLIndirectConvolutionLayer.h @@ -75,7 +75,12 @@ public: * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -90,7 +95,12 @@ public: * @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLIndirectConvolutionLayer * @@ -107,7 +117,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: @@ -117,5 +131,5 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLINDIRECTCONVOLUTIONLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h index 985a6a75f7..98d215dd4b 100644 --- a/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h @@ -83,7 +83,12 @@ public: * @param[in] epsilon (Optional) Lower bound value for the normalization. Defaults to 1e-12 * @param[in] use_mixed_precision (Optional) Use mixed precision in case of FP16 execution */ - void configure(ICLTensor *input, ICLTensor *output, float gamma = 1.0f, float beta = 0.0f, float epsilon = 1e-12f, bool use_mixed_precision = true); + void configure(ICLTensor *input, + ICLTensor *output, + float gamma = 1.0f, + float beta = 0.0f, + float epsilon = 1e-12f, + bool use_mixed_precision = true); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -95,7 +100,13 @@ public: * @param[in] epsilon (Optional) Lower bound value for the normalization. Defaults to 1e-12 * @param[in] use_mixed_precision (Optional) Use mixed precision in case of FP16 execution */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, float gamma = 1.0f, float beta = 0.0f, float epsilon = 1e-12f, bool use_mixed_precision = true); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + float gamma = 1.0f, + float beta = 0.0f, + float epsilon = 1e-12f, + bool use_mixed_precision = true); /** Static function to check if given info will lead to a valid configuration of @ref CLInstanceNormalizationLayer. * @@ -108,8 +119,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, float gamma = 1.0f, float beta = 0.0f, float epsilon = 1e-12f, bool use_mixed_precision = true); - void run() override; + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + float gamma = 1.0f, + float beta = 0.0f, + float epsilon = 1e-12f, + bool use_mixed_precision = true); + void run() override; private: std::unique_ptr<ICLKernel> _inst_norm_kernel; /**< Kernel to run */ diff --git a/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h b/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h index 4dc5c778d2..a8b356a708 100644 --- a/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h +++ b/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h @@ -26,8 +26,8 @@ #include "arm_compute/core/Types.h" #include "arm_compute/runtime/CL/CLTensor.h" -#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include "arm_compute/runtime/CL/functions/CLReductionOperation.h" +#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" @@ -89,7 +89,8 @@ public: * @param[in] axis Axis along which to reduce. Negative values wrap around. Maximum supported actual reduction axis : 2 * @param[in] epsilon (Optional) Lower bound value for the normalization. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, int axis, float epsilon = 1e-12f); + void configure( + const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, int axis, float epsilon = 1e-12f); /** Static function to check if given info will lead to a valid configuration of @ref CLL2NormalizeLayer. * @@ -111,5 +112,5 @@ private: std::unique_ptr<CLL2NormalizeLayerKernel> _normalize_kernel; CLTensor _sumsq; }; -} +} // namespace arm_compute #endif /*ARM_COMPUTE_CLL2NORMALIZELAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLLSTMLayer.h b/arm_compute/runtime/CL/functions/CLLSTMLayer.h index d26b4c5595..fe494991af 100644 --- a/arm_compute/runtime/CL/functions/CLLSTMLayer.h +++ b/arm_compute/runtime/CL/functions/CLLSTMLayer.h @@ -24,8 +24,6 @@ #ifndef ARM_COMPUTE_CLLSTMLAYER_H #define ARM_COMPUTE_CLLSTMLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" #include "arm_compute/runtime/CL/CLTensor.h" #include "arm_compute/runtime/CL/functions/CLActivationLayer.h" @@ -37,9 +35,10 @@ #include "arm_compute/runtime/CL/functions/CLGEMM.h" #include "arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h" #include "arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h" +#include "arm_compute/runtime/common/LSTMParams.h" +#include "arm_compute/runtime/IFunction.h" #include "arm_compute/runtime/IMemoryManager.h" #include "arm_compute/runtime/MemoryGroup.h" -#include "arm_compute/runtime/common/LSTMParams.h" #include <memory> @@ -53,7 +52,7 @@ namespace kernels { class ClTransposeKernel; } -} +} // namespace opencl /** This function performs a single time step in a Long Short-Term Memory (LSTM) layer. * @@ -120,13 +119,26 @@ public: * @param[in] projection_threshold (Optional) The clipping threshold for the output from the projection layer, such that values are bound within [-proj_clip, proj_clip]. * If set to 0.0f then clipping is disabled. */ - void configure(const ICLTensor *input, - const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - const ICLTensor *output_state_in, ICLTensor *cell_state_in, - ICLTensor *scratch_buffer, ICLTensor *output_state_out, ICLTensor *cell_state_out, ICLTensor *output, - const LSTMParams<ICLTensor> &lstm_params, const ActivationLayerInfo &activation_info, float cell_threshold = 0.f, float projection_threshold = 0.f); + void configure(const ICLTensor *input, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + const ICLTensor *output_state_in, + ICLTensor *cell_state_in, + ICLTensor *scratch_buffer, + ICLTensor *output_state_out, + ICLTensor *cell_state_out, + ICLTensor *output, + const LSTMParams<ICLTensor> &lstm_params, + const ActivationLayerInfo &activation_info, + float cell_threshold = 0.f, + float projection_threshold = 0.f); /** Initialize function's tensors. * * @param[in] compile_context The compile context to be used. @@ -166,13 +178,27 @@ public: * @param[in] projection_threshold (Optional) The clipping threshold for the output from the projection layer, such that values are bound within [-proj_clip, proj_clip]. * If set to 0.0f then clipping is disabled. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, - const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - const ICLTensor *output_state_in, ICLTensor *cell_state_in, - ICLTensor *scratch_buffer, ICLTensor *output_state_out, ICLTensor *cell_state_out, ICLTensor *output, - const LSTMParams<ICLTensor> &lstm_params, const ActivationLayerInfo &activation_info, float cell_threshold = 0.f, float projection_threshold = 0.f); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + const ICLTensor *output_state_in, + ICLTensor *cell_state_in, + ICLTensor *scratch_buffer, + ICLTensor *output_state_out, + ICLTensor *cell_state_out, + ICLTensor *output, + const LSTMParams<ICLTensor> &lstm_params, + const ActivationLayerInfo &activation_info, + float cell_threshold = 0.f, + float projection_threshold = 0.f); /** Static function to check if given info will lead to a valid configuration of @ref CLLSTMLayer * @@ -214,13 +240,26 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *output_state_in, const ITensorInfo *cell_state_in, - const ITensorInfo *scratch_buffer, const ITensorInfo *output_state_out, const ITensorInfo *cell_state_out, const ITensorInfo *output, - const LSTMParams<ITensorInfo> &lstm_params, const ActivationLayerInfo &activation_info, float cell_threshold = 0.f, float projection_threshold = 0.f); + static Status validate(const ITensorInfo *input, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_in, + const ITensorInfo *scratch_buffer, + const ITensorInfo *output_state_out, + const ITensorInfo *cell_state_out, + const ITensorInfo *output, + const LSTMParams<ITensorInfo> &lstm_params, + const ActivationLayerInfo &activation_info, + float cell_threshold = 0.f, + float projection_threshold = 0.f); // Inherited methods overridden: void run() override; @@ -311,7 +350,7 @@ private: bool _perform_projection_clipping; bool _is_prepared; bool _is_layer_norm_lstm; - const ICLTensor *_recurrent_to_cell_weights{ nullptr }; + const ICLTensor *_recurrent_to_cell_weights{nullptr}; }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLLSTMLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h b/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h index 9c004b85d0..8c116b1482 100644 --- a/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h +++ b/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h @@ -35,7 +35,6 @@ #include "arm_compute/runtime/CL/functions/CLQuantizationLayer.h" #include "arm_compute/runtime/CL/functions/CLSlice.h" #include "arm_compute/runtime/CL/functions/CLTranspose.h" - #include "arm_compute/runtime/common/LSTMParams.h" namespace arm_compute @@ -100,11 +99,22 @@ public: * @param[out] output_state_out Destination tensor. Output is a 2D tensor with dimensions [output_size, batch_size].Data types supported: Same as @p input. */ void configure(const ICLTensor *input, - const ICLTensor *input_to_input_weights, const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_input_weights, const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *input_gate_bias, const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - ICLTensor *cell_state_in, const ICLTensor *output_state_in, - ICLTensor *cell_state_out, ICLTensor *output_state_out); + const ICLTensor *input_to_input_weights, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_input_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *input_gate_bias, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + ICLTensor *cell_state_in, + const ICLTensor *output_state_in, + ICLTensor *cell_state_out, + ICLTensor *output_state_out); /** Initialize function's tensors. * * @param[in] compile_context The compile context to be used. @@ -126,12 +136,24 @@ public: * @param[out] cell_state_out Destination tensor. Output is a 2D tensor with dimensions [output_size, batch_size]. Data type supported: QSYMM16. * @param[out] output_state_out Destination tensor. Output is a 2D tensor with dimensions [output_size, batch_size].Data types supported: Same as @p input. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, - const ICLTensor *input_to_input_weights, const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_input_weights, const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *input_gate_bias, const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - ICLTensor *cell_state_in, const ICLTensor *output_state_in, - ICLTensor *cell_state_out, ICLTensor *output_state_out); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *input_to_input_weights, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_input_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *input_gate_bias, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + ICLTensor *cell_state_in, + const ICLTensor *output_state_in, + ICLTensor *cell_state_out, + ICLTensor *output_state_out); /** Static function to check if given info will lead to a valid configuration of @ref CLLSTMLayerQuantized * @@ -156,11 +178,22 @@ public: * @return a status */ static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_input_weights, const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_input_weights, const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *input_gate_bias, const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *cell_state_in, const ITensorInfo *output_state_in, - const ITensorInfo *cell_state_out, const ITensorInfo *output_state_out); + const ITensorInfo *input_to_input_weights, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_input_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *input_gate_bias, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *cell_state_in, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_out, + const ITensorInfo *output_state_out); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLLogicalAnd.h b/arm_compute/runtime/CL/functions/CLLogicalAnd.h index e3061e1dc3..4ff488782a 100644 --- a/arm_compute/runtime/CL/functions/CLLogicalAnd.h +++ b/arm_compute/runtime/CL/functions/CLLogicalAnd.h @@ -111,7 +111,8 @@ public: * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. * @param[out] output Output tensor. Data types supported: same as @p input1. */ - void configure(const CLCompileContext &compile_context, ITensorInfo *input1, ITensorInfo *input2, ITensorInfo *output); + void + configure(const CLCompileContext &compile_context, ITensorInfo *input1, ITensorInfo *input2, ITensorInfo *output); /** Static function to check if given info will lead to a valid configuration of @ref arm_compute::opencl::kernels::ClLogicalBinaryKernel * * @param[in] input1 First tensor input info. Data types supported: U8. diff --git a/arm_compute/runtime/CL/functions/CLLogicalNot.h b/arm_compute/runtime/CL/functions/CLLogicalNot.h index 27fd0f9c9f..c7d9db93d7 100644 --- a/arm_compute/runtime/CL/functions/CLLogicalNot.h +++ b/arm_compute/runtime/CL/functions/CLLogicalNot.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLLOGICALNOT_H #define ARM_COMPUTE_CLLOGICALNOT_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -85,7 +84,7 @@ public: * @return a status */ static Status validate(const ITensorInfo *input, const ITensorInfo *output); - void run() override; + void run() override; private: struct Impl; diff --git a/arm_compute/runtime/CL/functions/CLLogicalOr.h b/arm_compute/runtime/CL/functions/CLLogicalOr.h index 893c22f721..64b6d83177 100644 --- a/arm_compute/runtime/CL/functions/CLLogicalOr.h +++ b/arm_compute/runtime/CL/functions/CLLogicalOr.h @@ -111,7 +111,8 @@ public: * The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0. * @param[out] output Output tensor. Data types supported: same as @p input1. */ - void configure(const CLCompileContext &compile_context, ITensorInfo *input1, ITensorInfo *input2, ITensorInfo *output); + void + configure(const CLCompileContext &compile_context, ITensorInfo *input1, ITensorInfo *input2, ITensorInfo *output); /** Static function to check if given info will lead to a valid configuration of @ref arm_compute::opencl::kernels::ClLogicalBinaryKernel * * @param[in] input1 First tensor input info. Data types supported: U8. diff --git a/arm_compute/runtime/CL/functions/CLMatMul.h b/arm_compute/runtime/CL/functions/CLMatMul.h index 9d54bab868..9c9939b9d0 100644 --- a/arm_compute/runtime/CL/functions/CLMatMul.h +++ b/arm_compute/runtime/CL/functions/CLMatMul.h @@ -27,6 +27,7 @@ #include "arm_compute/core/Types.h" #include "arm_compute/function_info/ActivationLayerInfo.h" #include "arm_compute/runtime/IFunction.h" + #include <memory> namespace arm_compute @@ -88,14 +89,23 @@ public: * @param[in] settings Contains flags for function level settings * @param[in] act_info (Optional) Contains activation function and lower and upper bound values for bounded activation functions. */ - void configure(const CLCompileContext &compile_context, ICLTensor *rhs, ICLTensor *lhs, ICLTensor *dst, const MatMulInfo &matmul_info, const GpuMatMulSettings &settings = GpuMatMulSettings{}, const - ActivationLayerInfo &act_info = ActivationLayerInfo{}); + void configure(const CLCompileContext &compile_context, + ICLTensor *rhs, + ICLTensor *lhs, + ICLTensor *dst, + const MatMulInfo &matmul_info, + const GpuMatMulSettings &settings = GpuMatMulSettings{}, + const ActivationLayerInfo &act_info = ActivationLayerInfo{}); /** Initialise the kernel's inputs and output * * Similar to @ref CLMatMul::configure() */ - void configure(ICLTensor *lhs, ICLTensor *rhs, ICLTensor *dst, const MatMulInfo &matmul_info, const GpuMatMulSettings &settings = GpuMatMulSettings{}, const ActivationLayerInfo &act_info = - ActivationLayerInfo{}); + void configure(ICLTensor *lhs, + ICLTensor *rhs, + ICLTensor *dst, + const MatMulInfo &matmul_info, + const GpuMatMulSettings &settings = GpuMatMulSettings{}, + const ActivationLayerInfo &act_info = ActivationLayerInfo{}); /** Static function to check if given info will lead to a valid configuration of @ref CLMatMul. * * @@ -107,7 +117,11 @@ public: * @param[in] matmul_info Contains MatMul operation information described in @ref MatMulInfo. * @param[in] act_info (Optional) Contains activation function and lower and upper bound values for bounded activation functions. */ - static Status validate(const ITensorInfo *lhs, const ITensorInfo *rhs, const ITensorInfo *output, const MatMulInfo &matmul_info, const ActivationLayerInfo &act_info = ActivationLayerInfo{}); + static Status validate(const ITensorInfo *lhs, + const ITensorInfo *rhs, + const ITensorInfo *output, + const MatMulInfo &matmul_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo{}); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h b/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h index f7ff1234f6..2d2f064b4c 100644 --- a/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h +++ b/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h @@ -92,7 +92,11 @@ public: * @param[out] output Destination tensor. Data types supported: Same as @p input. * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *indices, ICLTensor *output, const PoolingLayerInfo &pool_info); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *indices, + ICLTensor *output, + const PoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLMaxUnpoolingLayer * * @param[in] input Source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. @@ -105,7 +109,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *indices, const ITensorInfo *output, const PoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *indices, + const ITensorInfo *output, + const PoolingLayerInfo &pool_info); // Inherited methods overridden: void run() override; @@ -114,5 +121,5 @@ private: CLFill _fill; std::unique_ptr<CLMaxUnpoolingLayerKernel> _unpooling_layer_kernel; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLMAXUNPOOLINGLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h index 68a7df24e6..951db3e419 100644 --- a/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h @@ -65,7 +65,10 @@ public: * @param[out] output (Optional) Destination tensor. It can be nullptr in case of in-place computation. Data type supported: same as @p input * @param[in] epsilon (Optional) Small float to avoid division by zero in case of zero standard deviation. Defaults to 1e-8. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output = nullptr, float epsilon = 1e-8f); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output = nullptr, + float epsilon = 1e-8f); /** Static function to check if given info will lead to a valid configuration of @ref CLMeanStdDevNormalizationKernel * * @param[in] input Source tensor info with 2 dimensions. In case of @p output tensor info = nullptr, diff --git a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h index 15406f7728..10fd8ed4c6 100644 --- a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h +++ b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h @@ -89,7 +89,10 @@ public: * Data types supported: same as @p input. Data layouts supported: same as @p input. * @param[in] norm_info Normalization layer information like the normalization type, normalization size and other parameters. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const NormalizationLayerInfo &norm_info); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + const NormalizationLayerInfo &norm_info); /** Static function to check if given info will lead to a valid configuration of @ref CLNormalizationLayer * * @param[in] input Source tensor. 3 lower dims represent a single input with dimensions [width, height, IFM], @@ -100,7 +103,8 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const NormalizationLayerInfo &norm_info); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const NormalizationLayerInfo &norm_info); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h b/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h index de5155c65a..3473af1004 100644 --- a/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h +++ b/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h @@ -62,7 +62,11 @@ public: * @param[in] std Standard deviation values tensor. 1 dimension with size equal to the number of input channels. * Data types supported: Same as @p input */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const ICLTensor *mean, const ICLTensor *std); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const ICLTensor *mean, + const ICLTensor *std); /** Static function to check if given info will lead to a valid configuration of @ref CLNormalizePlanarYUVLayer * * @param[in] input Source tensor info. 3 lower dimensions represent a single input with dimensions [width, height, channels]. @@ -74,7 +78,8 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *mean, const ITensorInfo *std); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *mean, const ITensorInfo *std); }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLNORMALIZEPLANARYUVLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLPadLayer.h b/arm_compute/runtime/CL/functions/CLPadLayer.h index 7f950bcfb3..89e693bd92 100644 --- a/arm_compute/runtime/CL/functions/CLPadLayer.h +++ b/arm_compute/runtime/CL/functions/CLPadLayer.h @@ -76,7 +76,11 @@ public: * @param[in] mode (Optional) Controls whether the padding should be filled with @p constant_value using CONSTANT, * or reflect the input, either including the border values (SYMMETRIC) or not (REFLECT). */ - void configure(ICLTensor *input, ICLTensor *output, const PaddingList &padding, PixelValue constant_value = PixelValue(), PaddingMode mode = PaddingMode::CONSTANT); + void configure(ICLTensor *input, + ICLTensor *output, + const PaddingList &padding, + PixelValue constant_value = PixelValue(), + PaddingMode mode = PaddingMode::CONSTANT); /** Initialize the function * * @param[in] compile_context The compile context to be used. @@ -88,8 +92,12 @@ public: * @param[in] mode (Optional) Controls whether the padding should be filled with @p constant_value using CONSTANT, * or reflect the input, either including the border values (SYMMETRIC) or not (REFLECT). */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const PaddingList &padding, PixelValue constant_value = PixelValue(), - PaddingMode mode = PaddingMode::CONSTANT); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + const PaddingList &padding, + PixelValue constant_value = PixelValue(), + PaddingMode mode = PaddingMode::CONSTANT); /** Static function to check if given info will lead to a valid configuration of @ref CLPadLayer. * @@ -101,7 +109,11 @@ public: * @param[in] mode (Optional) Controls whether the padding should be filled with @p constant_value using CONSTANT, * or reflect the input, either including the border values (SYMMETRIC) or not (REFLECT). */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PaddingList &padding, PixelValue constant_value = PixelValue(), PaddingMode mode = PaddingMode::CONSTANT); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const PaddingList &padding, + PixelValue constant_value = PixelValue(), + PaddingMode mode = PaddingMode::CONSTANT); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLPermute.h b/arm_compute/runtime/CL/functions/CLPermute.h index 8e15da2287..7ac0bf6b9c 100644 --- a/arm_compute/runtime/CL/functions/CLPermute.h +++ b/arm_compute/runtime/CL/functions/CLPermute.h @@ -78,7 +78,10 @@ public: * @param[in] output The output tensor. Data types supported: Same as @p input * @param[in] perm Permutation vector */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const PermutationVector &perm); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const PermutationVector &perm); /** Static function to check if given info will lead to a valid configuration of @ref CLPermute. * * @note Arbitrary permutation vectors are supported with rank not greater than 4 diff --git a/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h index 62b6d96ad5..f3e5cf9bd3 100644 --- a/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h +++ b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h @@ -84,8 +84,13 @@ public: * @param[in] rounding_policy Rounding policy. Supported rounding modes: to zero, to nearest even. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, float scale, - ConvertPolicy overflow_policy, RoundingPolicy rounding_policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + float scale, + ConvertPolicy overflow_policy, + RoundingPolicy rounding_policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output and convertion policy. * * @param[in] compile_context The compile context to be used. @@ -100,8 +105,14 @@ public: * @param[in] rounding_policy Rounding policy. Supported rounding modes: to zero, to nearest even. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, float scale, - ConvertPolicy overflow_policy, RoundingPolicy rounding_policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + float scale, + ConvertPolicy overflow_policy, + RoundingPolicy rounding_policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLPixelWiseMultiplication * * @param[in] input1 An input tensor info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/F16/F32. @@ -115,8 +126,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, float scale, - ConvertPolicy overflow_policy, RoundingPolicy rounding_policy, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + float scale, + ConvertPolicy overflow_policy, + RoundingPolicy rounding_policy, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; @@ -151,7 +167,10 @@ public: * @param[out] output The output tensor, Data types supported: same as @p input1. Number of channels supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Initialise the kernel's inputs, output. * * @param[in] compile_context The compile context to be used. @@ -162,7 +181,11 @@ public: * @param[out] output The output tensor, Data types supported: same as @p input1. Number of channels supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input1, ICLTensor *input2, ICLTensor *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + void configure(const CLCompileContext &compile_context, + ICLTensor *input1, + ICLTensor *input2, + ICLTensor *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); /** Static function to check if given info will lead to a valid configuration of @ref CLComplexPixelWiseMultiplication * * @param[in] input1 An input tensor info. Data types supported: F16/F32. Number of channels supported: 2. @@ -170,7 +193,10 @@ public: * @param[in] output The output tensor info, Data types supported: same as @p input1. Number of channels supported: same as @p input1. * @param[in] act_info (Optional) Activation layer information in case of a fused activation. */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ActivationLayerInfo &act_info = ActivationLayerInfo()); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const ActivationLayerInfo &act_info = ActivationLayerInfo()); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLPooling3dLayer.h b/arm_compute/runtime/CL/functions/CLPooling3dLayer.h index 91c46770da..1c69148771 100644 --- a/arm_compute/runtime/CL/functions/CLPooling3dLayer.h +++ b/arm_compute/runtime/CL/functions/CLPooling3dLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLPOOLING3DLAYER_H #define ARM_COMPUTE_CLPOOLING3DLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -82,7 +81,10 @@ public: * @param[out] output Destination tensor. Data types supported: Same as @p input. * @param[in] pool_info Contains 3d pooling operation information described in @ref Pooling3dLayerInfo. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const Pooling3dLayerInfo &pool_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const Pooling3dLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLPooling3dLayer * * @param[in] input Source tensor info. Data types supported: F16/F32/QASYMM8/QASYMM8_SIGNED. diff --git a/arm_compute/runtime/CL/functions/CLPoolingLayer.h b/arm_compute/runtime/CL/functions/CLPoolingLayer.h index 2163c16801..3dbdf8aeea 100644 --- a/arm_compute/runtime/CL/functions/CLPoolingLayer.h +++ b/arm_compute/runtime/CL/functions/CLPoolingLayer.h @@ -24,9 +24,8 @@ #ifndef ARM_COMPUTE_CLPOOLINGLAYER_H #define ARM_COMPUTE_CLPOOLINGLAYER_H -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/core/Types.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -74,7 +73,8 @@ public: * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. * @param[out] indices (optional) The indices of the maximal values. Data type supported: U32. */ - void configure(ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info, ICLTensor *indices = nullptr); + void + configure(ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info, ICLTensor *indices = nullptr); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -83,7 +83,11 @@ public: * @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo. * @param[out] indices (optional) The indices of the maximal values. Data type supported: U32. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info, ICLTensor *indices = nullptr); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + const PoolingLayerInfo &pool_info, + ICLTensor *indices = nullptr); /** Static function to check if given info will lead to a valid configuration of @ref CLPoolingLayer * * @param[in] input Source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. @@ -93,7 +97,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PoolingLayerInfo &pool_info, const ITensorInfo *indices = nullptr); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const PoolingLayerInfo &pool_info, + const ITensorInfo *indices = nullptr); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h b/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h index 9b36c9e433..4ede906baa 100644 --- a/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h +++ b/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h @@ -66,7 +66,11 @@ public: * @param[out] output Destination tensor. Output dimensions are [W * H * num_priors * 4, 2]. Data types and layouts supported: same as @p input1 * @param[in] info Prior box layer info. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, const PriorBoxLayerInfo &info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input1, + const ICLTensor *input2, + ICLTensor *output, + const PriorBoxLayerInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref CLPriorBoxLayer * * @param[in] input1 First source tensor info. Data types supported: F32. Data layouts supported: NCHW/NHWC. @@ -76,12 +80,15 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const PriorBoxLayerInfo &info); + static Status validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *output, + const PriorBoxLayerInfo &info); private: cl::Buffer _min; cl::Buffer _max; cl::Buffer _aspect_ratios; }; -} // arm_compute +} // namespace arm_compute #endif /* ARM_COMPUTE_CLPRIORBOXLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLQLSTMLayer.h b/arm_compute/runtime/CL/functions/CLQLSTMLayer.h index 1b0b759d74..3e76da086f 100644 --- a/arm_compute/runtime/CL/functions/CLQLSTMLayer.h +++ b/arm_compute/runtime/CL/functions/CLQLSTMLayer.h @@ -32,7 +32,6 @@ #include "arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h" #include "arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h" #include "arm_compute/runtime/CL/functions/CLTranspose.h" - #include "arm_compute/runtime/common/LSTMParams.h" namespace arm_compute @@ -127,12 +126,21 @@ public: * projection_threshold (Optional) The clipping threshold for the output from the projection layer, such that values are bound within * [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled. */ - void configure(const ICLTensor *input, - const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - ICLTensor *cell_state_in, ICLTensor *output_state_in, - ICLTensor *cell_state_out, ICLTensor *output_state_out, ICLTensor *output, + void configure(const ICLTensor *input, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + ICLTensor *cell_state_in, + ICLTensor *output_state_in, + ICLTensor *cell_state_out, + ICLTensor *output_state_out, + ICLTensor *output, const LSTMParams<ICLTensor> &lstm_params); /** Initialize function's tensors. @@ -177,12 +185,22 @@ public: * projection_threshold (Optional) The clipping threshold for the output from the projection layer, such that values are bound within * [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, - const ICLTensor *input_to_forget_weights, const ICLTensor *input_to_cell_weights, const ICLTensor *input_to_output_weights, - const ICLTensor *recurrent_to_forget_weights, const ICLTensor *recurrent_to_cell_weights, const ICLTensor *recurrent_to_output_weights, - const ICLTensor *forget_gate_bias, const ICLTensor *cell_bias, const ICLTensor *output_gate_bias, - ICLTensor *cell_state_in, ICLTensor *output_state_in, - ICLTensor *cell_state_out, ICLTensor *output_state_out, ICLTensor *output, + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *input_to_forget_weights, + const ICLTensor *input_to_cell_weights, + const ICLTensor *input_to_output_weights, + const ICLTensor *recurrent_to_forget_weights, + const ICLTensor *recurrent_to_cell_weights, + const ICLTensor *recurrent_to_output_weights, + const ICLTensor *forget_gate_bias, + const ICLTensor *cell_bias, + const ICLTensor *output_gate_bias, + ICLTensor *cell_state_in, + ICLTensor *output_state_in, + ICLTensor *cell_state_out, + ICLTensor *output_state_out, + ICLTensor *output, const LSTMParams<ICLTensor> &lstm_params); /** Static function to check if given info will lead to a valid configuration of @ref CLQLSTMLayer @@ -227,12 +245,21 @@ public: * [-proj_clip, proj_clip]. If set to 0.0 then clipping is disabled. * @return a status */ - static Status validate(const ITensorInfo *input, - const ITensorInfo *input_to_forget_weights, const ITensorInfo *input_to_cell_weights, const ITensorInfo *input_to_output_weights, - const ITensorInfo *recurrent_to_forget_weights, const ITensorInfo *recurrent_to_cell_weights, const ITensorInfo *recurrent_to_output_weights, - const ITensorInfo *forget_gate_bias, const ITensorInfo *cell_bias, const ITensorInfo *output_gate_bias, - const ITensorInfo *cell_state_in, const ITensorInfo *output_state_in, - const ITensorInfo *cell_state_out, const ITensorInfo *output_state_out, const ITensorInfo *output, + static Status validate(const ITensorInfo *input, + const ITensorInfo *input_to_forget_weights, + const ITensorInfo *input_to_cell_weights, + const ITensorInfo *input_to_output_weights, + const ITensorInfo *recurrent_to_forget_weights, + const ITensorInfo *recurrent_to_cell_weights, + const ITensorInfo *recurrent_to_output_weights, + const ITensorInfo *forget_gate_bias, + const ITensorInfo *cell_bias, + const ITensorInfo *output_gate_bias, + const ITensorInfo *cell_state_in, + const ITensorInfo *output_state_in, + const ITensorInfo *cell_state_out, + const ITensorInfo *output_state_out, + const ITensorInfo *output, const LSTMParams<ITensorInfo> &lstm_params); // Inherited methods overridden: @@ -266,10 +293,18 @@ private: * @param[in] mm_res_info Tensor info to be used to initialize output stage result tensor. * */ - void configure_mm(const CLCompileContext &compile_context, CLGEMMLowpMatrixMultiplyCore &mm, CLGEMMLowpOutputStage &outstage, GEMMLowpOutputStageInfo &gemmlowp_info, - const ICLTensor *mm_input, const ICLTensor *mm_weights, const ICLTensor *bias, CLTensor *mm_res, - CLTensor *outstage_res, float gemmlowp_scale, - const TensorInfo &mm_res_info, const TensorInfo &outstage_tensor_info); + void configure_mm(const CLCompileContext &compile_context, + CLGEMMLowpMatrixMultiplyCore &mm, + CLGEMMLowpOutputStage &outstage, + GEMMLowpOutputStageInfo &gemmlowp_info, + const ICLTensor *mm_input, + const ICLTensor *mm_weights, + const ICLTensor *bias, + CLTensor *mm_res, + CLTensor *outstage_res, + float gemmlowp_scale, + const TensorInfo &mm_res_info, + const TensorInfo &outstage_tensor_info); MemoryGroup _memory_group{}; @@ -278,8 +313,8 @@ private: { static constexpr uint32_t max_dimension_supported = 2; - ICLTensor *_src{ nullptr }; - ICLTensor *_dst{ nullptr }; + ICLTensor *_src{nullptr}; + ICLTensor *_dst{nullptr}; size_t _row_size{}; Window _window{}; @@ -368,7 +403,7 @@ private: CLArithmeticAddition _accumulate_projection{}; CLActivationLayer _projection_clip{}; std::array<std::unique_ptr<CLQLSTMLayerNormalizationKernel>, _layer_norm_count> _layer_norms; - CLCopy _copy_output; + CLCopy _copy_output; TensorCopyKernel _projection_bias_copy{}; TensorCopyKernel _projection_output_to_accumulate_copy{}; @@ -376,21 +411,18 @@ private: TensorCopyKernel _hidden_to_output_copy{}; // Tensor pointers - const ICLTensor *_input_to_input_weights - { - nullptr - }; - const ICLTensor *_recurrent_to_input_weights{ nullptr }; - const ICLTensor *_projection_bias{ nullptr }; - const ICLTensor *_input_to_forget_weights{ nullptr }; - const ICLTensor *_input_to_cell_weights{ nullptr }; - const ICLTensor *_input_to_output_weights{ nullptr }; - const ICLTensor *_recurrent_to_forget_weights{ nullptr }; - const ICLTensor *_recurrent_to_cell_weights{ nullptr }; - const ICLTensor *_recurrent_to_output_weights{ nullptr }; - const ICLTensor *_projection_weights{ nullptr }; - std::array<const ICLTensor *, _layer_norm_count> _layer_norm_weights{ {} }; - std::array<const ICLTensor *, _layer_norm_count> _layer_norm_bias{ {} }; + const ICLTensor *_input_to_input_weights{nullptr}; + const ICLTensor *_recurrent_to_input_weights{nullptr}; + const ICLTensor *_projection_bias{nullptr}; + const ICLTensor *_input_to_forget_weights{nullptr}; + const ICLTensor *_input_to_cell_weights{nullptr}; + const ICLTensor *_input_to_output_weights{nullptr}; + const ICLTensor *_recurrent_to_forget_weights{nullptr}; + const ICLTensor *_recurrent_to_cell_weights{nullptr}; + const ICLTensor *_recurrent_to_output_weights{nullptr}; + const ICLTensor *_projection_weights{nullptr}; + std::array<const ICLTensor *, _layer_norm_count> _layer_norm_weights{{}}; + std::array<const ICLTensor *, _layer_norm_count> _layer_norm_bias{{}}; using LayerNormIndexType = typename std::underlying_type<LayerNormGate>::type; inline LayerNormIndexType getGateIndex(LayerNormGate g) @@ -423,78 +455,78 @@ private: return *_layer_norms[getGateIndex(g)]; } - inline void configure_layer_norm(LayerNormGate g, const ICLTensor *in); + inline void configure_layer_norm(LayerNormGate g, const ICLTensor *in); inline static Status validate_layer_norm(const ITensorInfo &in, const ITensorInfo &weight, const ITensorInfo &bias); // Temporary tensors - CLTensor _input_to_forget_weights_transposed{ nullptr }; - CLTensor _input_to_cell_weights_transposed{ nullptr }; - CLTensor _input_to_output_weights_transposed{ nullptr }; - CLTensor _input_to_input_weights_transposed{ nullptr }; - CLTensor _recurrent_to_forget_weights_transposed{ nullptr }; - CLTensor _recurrent_to_cell_weights_transposed{ nullptr }; - CLTensor _recurrent_to_output_weights_transposed{ nullptr }; - CLTensor _recurrent_to_input_weights_transposed{ nullptr }; - CLTensor _projection_weights_transposed{ nullptr }; - CLTensor _input_to_input_eff_bias{ nullptr }; - CLTensor _recurrent_to_input_eff_bias{ nullptr }; - CLTensor _input_to_forget_eff_bias{ nullptr }; - CLTensor _recurrent_to_forget_eff_bias{ nullptr }; - CLTensor _input_to_cell_eff_bias{ nullptr }; - CLTensor _recurrent_to_cell_eff_bias{ nullptr }; - CLTensor _input_to_output_eff_bias{ nullptr }; - CLTensor _recurrent_to_output_eff_bias{ nullptr }; - CLTensor _projection_reduction_res{ nullptr }; - CLTensor _projection_eff_bias{ nullptr }; - CLTensor _mm_input_to_forget_res{ nullptr }; - CLTensor _mm_recurrent_to_forget_res{ nullptr }; - CLTensor _mul_cell_to_forget_res{ nullptr }; - CLTensor _input_to_forget_outstage_res{ nullptr }; - CLTensor _cell_to_forget_outstage_res{ nullptr }; - CLTensor _recurrent_to_forget_outstage_res{ nullptr }; - CLTensor _forget_gate{ nullptr }; - CLTensor _mm_input_to_cell_res{ nullptr }; - CLTensor _input_to_cell_outstage_res{ nullptr }; - CLTensor _mm_recurrent_to_cell_res{ nullptr }; - CLTensor _recurrent_to_cell_outstage_res{ nullptr }; - CLTensor _cell_gate{ nullptr }; - CLTensor _mul_input_cell_res{ nullptr }; - CLTensor _mm_input_to_input_res{ nullptr }; - CLTensor _input_to_input_outstage_res{ nullptr }; - CLTensor _mm_recurrent_to_input_res{ nullptr }; - CLTensor _mul_cell_to_input_res{ nullptr }; - CLTensor _cell_to_input_outstage_res{ nullptr }; - CLTensor _recurrent_to_input_outstage_res{ nullptr }; - CLTensor _input_gate{ nullptr }; - CLTensor _mm_input_to_output_res{ nullptr }; - CLTensor _input_to_output_outstage_res{ nullptr }; - CLTensor _mm_recurrent_to_output_res{ nullptr }; - CLTensor _mul_cell_to_output_res{ nullptr }; - CLTensor _cell_to_output_outstage_res{ nullptr }; - CLTensor _recurrent_to_output_outstage_res{ nullptr }; - CLTensor _output_gate{ nullptr }; - CLTensor _hidden_mul_res{ nullptr }; - CLTensor _hidden_gate{ nullptr }; - CLTensor _mm_projection_res{ nullptr }; - CLTensor _projection_outstage_res{ nullptr }; - CLTensor _projection_out_res{ nullptr }; - CLTensor _projection_accumulate_res{ nullptr }; - CLTensor _ones{ nullptr }; - std::array<CLTensor, _layer_norm_count> _layer_norm_output{ {} }; + CLTensor _input_to_forget_weights_transposed{nullptr}; + CLTensor _input_to_cell_weights_transposed{nullptr}; + CLTensor _input_to_output_weights_transposed{nullptr}; + CLTensor _input_to_input_weights_transposed{nullptr}; + CLTensor _recurrent_to_forget_weights_transposed{nullptr}; + CLTensor _recurrent_to_cell_weights_transposed{nullptr}; + CLTensor _recurrent_to_output_weights_transposed{nullptr}; + CLTensor _recurrent_to_input_weights_transposed{nullptr}; + CLTensor _projection_weights_transposed{nullptr}; + CLTensor _input_to_input_eff_bias{nullptr}; + CLTensor _recurrent_to_input_eff_bias{nullptr}; + CLTensor _input_to_forget_eff_bias{nullptr}; + CLTensor _recurrent_to_forget_eff_bias{nullptr}; + CLTensor _input_to_cell_eff_bias{nullptr}; + CLTensor _recurrent_to_cell_eff_bias{nullptr}; + CLTensor _input_to_output_eff_bias{nullptr}; + CLTensor _recurrent_to_output_eff_bias{nullptr}; + CLTensor _projection_reduction_res{nullptr}; + CLTensor _projection_eff_bias{nullptr}; + CLTensor _mm_input_to_forget_res{nullptr}; + CLTensor _mm_recurrent_to_forget_res{nullptr}; + CLTensor _mul_cell_to_forget_res{nullptr}; + CLTensor _input_to_forget_outstage_res{nullptr}; + CLTensor _cell_to_forget_outstage_res{nullptr}; + CLTensor _recurrent_to_forget_outstage_res{nullptr}; + CLTensor _forget_gate{nullptr}; + CLTensor _mm_input_to_cell_res{nullptr}; + CLTensor _input_to_cell_outstage_res{nullptr}; + CLTensor _mm_recurrent_to_cell_res{nullptr}; + CLTensor _recurrent_to_cell_outstage_res{nullptr}; + CLTensor _cell_gate{nullptr}; + CLTensor _mul_input_cell_res{nullptr}; + CLTensor _mm_input_to_input_res{nullptr}; + CLTensor _input_to_input_outstage_res{nullptr}; + CLTensor _mm_recurrent_to_input_res{nullptr}; + CLTensor _mul_cell_to_input_res{nullptr}; + CLTensor _cell_to_input_outstage_res{nullptr}; + CLTensor _recurrent_to_input_outstage_res{nullptr}; + CLTensor _input_gate{nullptr}; + CLTensor _mm_input_to_output_res{nullptr}; + CLTensor _input_to_output_outstage_res{nullptr}; + CLTensor _mm_recurrent_to_output_res{nullptr}; + CLTensor _mul_cell_to_output_res{nullptr}; + CLTensor _cell_to_output_outstage_res{nullptr}; + CLTensor _recurrent_to_output_outstage_res{nullptr}; + CLTensor _output_gate{nullptr}; + CLTensor _hidden_mul_res{nullptr}; + CLTensor _hidden_gate{nullptr}; + CLTensor _mm_projection_res{nullptr}; + CLTensor _projection_outstage_res{nullptr}; + CLTensor _projection_out_res{nullptr}; + CLTensor _projection_accumulate_res{nullptr}; + CLTensor _ones{nullptr}; + std::array<CLTensor, _layer_norm_count> _layer_norm_output{{}}; inline CLTensor &get_layer_norm_output(LayerNormGate g) { return _layer_norm_output[getGateIndex(g)]; } - bool _is_prepared{ false }; - bool _has_cifg{ false }; - bool _has_cell_clipping{ false }; - bool _has_projection{ false }; - bool _has_projection_clipping{ false }; - bool _has_peephole{ false }; - bool _has_layer_norm{ false }; - bool _projection_tensor_copy_required{ false }; + bool _is_prepared{false}; + bool _has_cifg{false}; + bool _has_cell_clipping{false}; + bool _has_projection{false}; + bool _has_projection_clipping{false}; + bool _has_peephole{false}; + bool _has_layer_norm{false}; + bool _projection_tensor_copy_required{false}; }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLQLSTMLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLRNNLayer.h b/arm_compute/runtime/CL/functions/CLRNNLayer.h index 2b3b35e37d..a8d835d04d 100644 --- a/arm_compute/runtime/CL/functions/CLRNNLayer.h +++ b/arm_compute/runtime/CL/functions/CLRNNLayer.h @@ -24,12 +24,12 @@ #ifndef ARM_COMPUTE_CLRNN_LAYER_H #define ARM_COMPUTE_CLRNN_LAYER_H -#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include "arm_compute/runtime/CL/functions/CLActivationLayer.h" #include "arm_compute/runtime/CL/functions/CLCopy.h" #include "arm_compute/runtime/CL/functions/CLElementwiseOperations.h" #include "arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h" #include "arm_compute/runtime/CL/functions/CLGEMM.h" +#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include <memory> @@ -69,7 +69,13 @@ public: * @param[in,out] hidden_state Output tensor of shape [num_units, batch_size]. Data types supported: Same as @p input * @param[in] info Activation layer parameter. */ - void configure(const ICLTensor *input, const ICLTensor *weights, const ICLTensor *recurrent_weights, const ICLTensor *bias, ICLTensor *hidden_state, ICLTensor *output, ActivationLayerInfo &info); + void configure(const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *recurrent_weights, + const ICLTensor *bias, + ICLTensor *hidden_state, + ICLTensor *output, + ActivationLayerInfo &info); /** Initialize the function * * @param[in] compile_context The compile context to be used. @@ -81,8 +87,14 @@ public: * @param[in,out] hidden_state Output tensor of shape [num_units, batch_size]. Data types supported: Same as @p input * @param[in] info Activation layer parameter. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *weights, const ICLTensor *recurrent_weights, const ICLTensor *bias, ICLTensor *hidden_state, - ICLTensor *output, ActivationLayerInfo &info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *recurrent_weights, + const ICLTensor *bias, + ICLTensor *hidden_state, + ICLTensor *output, + ActivationLayerInfo &info); /** Initialize the function * * @param[in] input Input is a 2-D tensor of shape [input_size, batch_size]. Data types supported: F16/F32 @@ -95,7 +107,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *recurrent_weights, const ITensorInfo *bias, const ITensorInfo *hidden_state, const ITensorInfo *output, + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *recurrent_weights, + const ITensorInfo *bias, + const ITensorInfo *hidden_state, + const ITensorInfo *output, const ActivationLayerInfo &info); // Inherited methods overridden: @@ -114,5 +131,5 @@ private: CLTensor _add_output; bool _is_prepared; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLRNN_LAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLROIAlignLayer.h b/arm_compute/runtime/CL/functions/CLROIAlignLayer.h index 1eaea1b297..14d3476711 100644 --- a/arm_compute/runtime/CL/functions/CLROIAlignLayer.h +++ b/arm_compute/runtime/CL/functions/CLROIAlignLayer.h @@ -68,7 +68,8 @@ public: * @note The z dimensions of @p output tensor and @p input tensor must be the same. * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. */ - void configure(const ICLTensor *input, const ICLTensor *rois, ICLTensor *output, const ROIPoolingLayerInfo &pool_info); + void + configure(const ICLTensor *input, const ICLTensor *rois, ICLTensor *output, const ROIPoolingLayerInfo &pool_info); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -84,7 +85,11 @@ public: * @note The z dimensions of @p output tensor and @p input tensor must be the same. * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *rois, ICLTensor *output, const ROIPoolingLayerInfo &pool_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *rois, + ICLTensor *output, + const ROIPoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLROIAlignLayer * * @param[in] input Source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. @@ -100,7 +105,10 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *rois, ITensorInfo *output, const ROIPoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *rois, + ITensorInfo *output, + const ROIPoolingLayerInfo &pool_info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLROIALIGNLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h b/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h index 151586a1f6..86294596d2 100644 --- a/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h +++ b/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h @@ -66,7 +66,8 @@ public: * @note The z dimensions of @p output tensor and @p input tensor must be the same. * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. */ - void configure(const ICLTensor *input, const ICLTensor *rois, ICLTensor *output, const ROIPoolingLayerInfo &pool_info); + void + configure(const ICLTensor *input, const ICLTensor *rois, ICLTensor *output, const ROIPoolingLayerInfo &pool_info); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -81,7 +82,11 @@ public: * @note The z dimensions of @p output tensor and @p input tensor must be the same. * @note The fourth dimension of @p output tensor must be the same as the number of elements in @p rois array. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *rois, const ICLTensor *output, const ROIPoolingLayerInfo &pool_info); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *rois, + const ICLTensor *output, + const ROIPoolingLayerInfo &pool_info); /** Static function to check if given info will lead to a valid configuration of @ref CLROIPoolingLayer * @@ -97,7 +102,10 @@ public: * * @return a Status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *rois, ITensorInfo *output, const ROIPoolingLayerInfo &pool_info); + static Status validate(const ITensorInfo *input, + const ITensorInfo *rois, + ITensorInfo *output, + const ROIPoolingLayerInfo &pool_info); }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLROIPOOLINGLAYER_H */ diff --git a/arm_compute/runtime/CL/functions/CLRange.h b/arm_compute/runtime/CL/functions/CLRange.h index fbce05162c..ed665bc398 100644 --- a/arm_compute/runtime/CL/functions/CLRange.h +++ b/arm_compute/runtime/CL/functions/CLRange.h @@ -73,7 +73,8 @@ public: * @param[in] end The ending (not including) value of the sequence. * @param[in] step The gap between each pair of values in the sequence. Default is 1. */ - void configure(const CLCompileContext &compile_context, ICLTensor *output, float start, float end, float step = 1.f); + void + configure(const CLCompileContext &compile_context, ICLTensor *output, float start, float end, float step = 1.f); /** Static function to check if given info will lead to a valid configuration of @ref CLRange * * @param[in] output Output tensor info. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32. diff --git a/arm_compute/runtime/CL/functions/CLReduceMean.h b/arm_compute/runtime/CL/functions/CLReduceMean.h index 1ce088b2ce..640fe7cf1b 100644 --- a/arm_compute/runtime/CL/functions/CLReduceMean.h +++ b/arm_compute/runtime/CL/functions/CLReduceMean.h @@ -24,12 +24,12 @@ #ifndef ARM_COMPUTE_CL_REDUCE_MEAN_H #define ARM_COMPUTE_CL_REDUCE_MEAN_H -#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include "arm_compute/runtime/CL/functions/CLDequantizationLayer.h" #include "arm_compute/runtime/CL/functions/CLElementwiseOperations.h" #include "arm_compute/runtime/CL/functions/CLQuantizationLayer.h" #include "arm_compute/runtime/CL/functions/CLReductionOperation.h" #include "arm_compute/runtime/CL/functions/CLReshapeLayer.h" +#include "arm_compute/runtime/CL/ICLSimpleFunction.h" #include "arm_compute/runtime/IMemoryManager.h" namespace arm_compute @@ -74,7 +74,11 @@ public: * @param[in] keep_dims If positive, retains reduced dimensions with length 1. * @param[out] output Destination tensor. Data type supported: Same as @p input */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const Coordinates &reduction_axis, bool keep_dims, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const Coordinates &reduction_axis, + bool keep_dims, + ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLReduceMean * @@ -85,7 +89,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output); + static Status + validate(const ITensorInfo *input, const Coordinates &reduction_axis, bool keep_dims, const ITensorInfo *output); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLReductionOperation.h b/arm_compute/runtime/CL/functions/CLReductionOperation.h index 2245735b62..80068ac35c 100644 --- a/arm_compute/runtime/CL/functions/CLReductionOperation.h +++ b/arm_compute/runtime/CL/functions/CLReductionOperation.h @@ -80,7 +80,8 @@ public: * @param[in] op Reduction operation to perform. Operations supported: MEAN_SUM, PROD, SUM_SQUARE, SUM, MIN, MAX * @param[in] keep_dims (Optional) Whether to keep the reduced dimension after the operation. Defaults to true. */ - void configure(ICLTensor *input, ICLTensor *output, unsigned int axis, ReductionOperation op, bool keep_dims = true); + void + configure(ICLTensor *input, ICLTensor *output, unsigned int axis, ReductionOperation op, bool keep_dims = true); /** Set the input and output tensors. * * @param[in] compile_context The compile context to be used. @@ -90,7 +91,12 @@ public: * @param[in] op Reduction operation to perform. Operations supported: MEAN_SUM, PROD, SUM_SQUARE, SUM, MIN, MAX * @param[in] keep_dims (Optional) Whether to keep the reduced dimension after the operation. Defaults to true. */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, unsigned int axis, ReductionOperation op, bool keep_dims = true); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + unsigned int axis, + ReductionOperation op, + bool keep_dims = true); /** Static function to check if given info will lead to a valid configuration of @ref CLReductionOperation. * @@ -102,7 +108,11 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, unsigned int axis, ReductionOperation op, bool keep_dims = true); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + unsigned int axis, + ReductionOperation op, + bool keep_dims = true); // Inherited methods overridden: void run() override; @@ -118,4 +128,4 @@ private: bool _is_reshape_required; }; } // namespace arm_compute -#endif /* ARM_COMPUTE_CLREDUCTIONOPERATION_H */
\ No newline at end of file +#endif /* ARM_COMPUTE_CLREDUCTIONOPERATION_H */ diff --git a/arm_compute/runtime/CL/functions/CLReshapeLayer.h b/arm_compute/runtime/CL/functions/CLReshapeLayer.h index 7346b65e9b..dad90e6ba9 100644 --- a/arm_compute/runtime/CL/functions/CLReshapeLayer.h +++ b/arm_compute/runtime/CL/functions/CLReshapeLayer.h @@ -26,6 +26,7 @@ #include "arm_compute/runtime/CL/ICLOperator.h" #include "arm_compute/runtime/CL/ICLSimpleFunction.h" + #include <memory> namespace arm_compute diff --git a/arm_compute/runtime/CL/functions/CLReverse.h b/arm_compute/runtime/CL/functions/CLReverse.h index 94c63ca92d..0defc3f28b 100644 --- a/arm_compute/runtime/CL/functions/CLReverse.h +++ b/arm_compute/runtime/CL/functions/CLReverse.h @@ -59,7 +59,10 @@ public: * @param[out] output Output tensor. Data type supported: Same as @p input * @param[in] axis Axis tensor. Contains the indices of the dimensions to reverse. Data type supported: U32 */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const ICLTensor *axis); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const ICLTensor *axis); /** Static function to check if given info will lead to a valid configuration of @ref CLReverseKernel * * @param[in] input Input tensor info. Data types supported: All. diff --git a/arm_compute/runtime/CL/functions/CLScale.h b/arm_compute/runtime/CL/functions/CLScale.h index ddb4a23531..5c3824eb58 100644 --- a/arm_compute/runtime/CL/functions/CLScale.h +++ b/arm_compute/runtime/CL/functions/CLScale.h @@ -83,7 +83,10 @@ public: * All but the lowest two dimensions must be the same size as in the input tensor, i.e. scaling is only performed within the XY-plane. * @param[in] info @ref ScaleKernelInfo descriptor to be used to configure */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, ICLTensor *output, const ScaleKernelInfo &info); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + ICLTensor *output, + const ScaleKernelInfo &info); /** Static function to check if given info will lead to a valid configuration of @ref CLScale * diff --git a/arm_compute/runtime/CL/functions/CLSelect.h b/arm_compute/runtime/CL/functions/CLSelect.h index 8b1e6b2019..effcb58313 100644 --- a/arm_compute/runtime/CL/functions/CLSelect.h +++ b/arm_compute/runtime/CL/functions/CLSelect.h @@ -62,7 +62,11 @@ public: * @param[in] y Second input tensor. Data types supported: Same as @p x * @param[out] output Output tensor. Data types supported: Same as @p x. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *c, const ICLTensor *x, const ICLTensor *y, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *c, + const ICLTensor *x, + const ICLTensor *y, + ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLSelect * * @param[in] c Condition input tensor. Data types supported: U8. diff --git a/arm_compute/runtime/CL/functions/CLSlice.h b/arm_compute/runtime/CL/functions/CLSlice.h index 297bcd86fe..7a274ded72 100644 --- a/arm_compute/runtime/CL/functions/CLSlice.h +++ b/arm_compute/runtime/CL/functions/CLSlice.h @@ -84,7 +84,11 @@ public: * @param[in] starts The starts of the dimensions of the input tensor to be sliced. The length must be of rank(input). * @param[in] ends The ends of the dimensions of the input tensor to be sliced. The length must be of rank(input). */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const Coordinates &starts, const Coordinates &ends); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const Coordinates &starts, + const Coordinates &ends); /** Static function to check if given info will lead to a valid configuration of @ref CLSlice * @@ -100,7 +104,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); // Inherited methods overridden: void run() override; @@ -129,7 +134,11 @@ public: * @param[in] starts The starts of the dimensions of the input tensor to be sliced. The length must be of rank(input). * @param[in] ends The ends of the dimensions of the input tensor to be sliced. The length must be of rank(input). */ - void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); + void configure(const CLCompileContext &compile_context, + const ITensorInfo *input, + ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends); /** Static function to check if given info will lead to a valid configuration of @ref CLSlice * @@ -145,7 +154,8 @@ public: * * @return A status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); + static Status + validate(const ITensorInfo *input, const ITensorInfo *output, const Coordinates &starts, const Coordinates &ends); }; } // namespace experimental } // namespace arm_compute diff --git a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h index 687f8ff6d8..70ef1f4402 100644 --- a/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h +++ b/arm_compute/runtime/CL/functions/CLSoftmaxLayer.h @@ -87,7 +87,11 @@ public: * @param[in] axis (Optional) The dimension in which to apply the function. E.g. for input of shape 4x5x6 and * axis=1, softmax will be applied to 4x6=24 vectors of size 5. Defaults to 0 */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, float beta = 1.0f, int32_t axis = 0); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + float beta = 1.0f, + int32_t axis = 0); /** Static function to check if given info will lead to a valid configuration of @ref CLSoftmaxLayer * * @param[in] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32 for Softmax and F16/F32 for Log Softmax diff --git a/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h b/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h index 304a74137e..191f4863d5 100644 --- a/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h +++ b/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h @@ -83,7 +83,11 @@ public: * @param[in] paddings 2-D tensor with shape [2, M] (First dimension is the fastest-changing dimension). Supported M: 2. Data types supported: S32 * @param[out] output Tensor output. Data types supported: same as @p input */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const ICLTensor *block_shape, const ICLTensor *paddings, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const ICLTensor *block_shape, + const ICLTensor *paddings, + ICLTensor *output); /** Set the input and output tensors. (Static block shape and paddings) * * @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: All. @@ -93,7 +97,12 @@ public: * @param[in] padding_right The padding at the end of every dimension of the output tensor. * @param[out] output Tensor output. Data types supported: same as @p input */ - void configure(const ICLTensor *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right, ICLTensor *output); + void configure(const ICLTensor *input, + const int block_shape_x, + const int block_shape_y, + const Size2D &padding_left, + const Size2D &padding_right, + ICLTensor *output); /** Set the input and output tensors. (Static block shape and paddings) * * @param[in] compile_context The compile context to be used. @@ -104,8 +113,13 @@ public: * @param[in] padding_right The padding at the end of every dimension of the output tensor. * @param[out] output Tensor output. Data types supported: same as @p input */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right, - ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const int block_shape_x, + const int block_shape_y, + const Size2D &padding_left, + const Size2D &padding_right, + ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLSpaceToBatchLayer * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -115,7 +129,10 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *paddings, const ITensorInfo *output); + static Status validate(const ITensorInfo *input, + const ITensorInfo *block_shape, + const ITensorInfo *paddings, + const ITensorInfo *output); /** Static function to check if given info will lead to a valid configuration of @ref CLSpaceToBatchLayer (Static block shape and paddings) * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. @@ -127,7 +144,12 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const int block_shape_x, const int block_shape_y, const Size2D &padding_left, const Size2D &padding_right, const ITensorInfo *output); + static Status validate(const ITensorInfo *input, + const int block_shape_x, + const int block_shape_y, + const Size2D &padding_left, + const Size2D &padding_right, + const ITensorInfo *output); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h b/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h index 8a47e95f9d..1b0dfc2b74 100644 --- a/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h +++ b/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h @@ -75,7 +75,8 @@ public: * @param[out] output Tensor output. Data types supported: same as @p input * @param[in] block_shape Block shape value. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, int32_t block_shape); + void + configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, int32_t block_shape); /** Static function to check if given info will lead to a valid configuration of @ref CLSpaceToDepthLayer. * * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All. diff --git a/arm_compute/runtime/CL/functions/CLSplit.h b/arm_compute/runtime/CL/functions/CLSplit.h index 86c7bdde7d..8d13755212 100644 --- a/arm_compute/runtime/CL/functions/CLSplit.h +++ b/arm_compute/runtime/CL/functions/CLSplit.h @@ -26,7 +26,6 @@ #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/Types.h" - #include "arm_compute/runtime/CL/functions/CLSlice.h" #include "arm_compute/runtime/CPP/functions/CPPSplit.h" #include "arm_compute/runtime/IFunction.h" diff --git a/arm_compute/runtime/CL/functions/CLStackLayer.h b/arm_compute/runtime/CL/functions/CLStackLayer.h index 54c903a706..18745c8a4f 100644 --- a/arm_compute/runtime/CL/functions/CLStackLayer.h +++ b/arm_compute/runtime/CL/functions/CLStackLayer.h @@ -85,7 +85,10 @@ public: * Negative values wrap around * @param[out] output Output tensor. Data types supported: Same as @p input. */ - void configure(const CLCompileContext &compile_context, const std::vector<ICLTensor *> &input, int axis, ICLTensor *output); + void configure(const CLCompileContext &compile_context, + const std::vector<ICLTensor *> &input, + int axis, + ICLTensor *output); /** Static function to check if given info will lead to a valid configuration of @ref CLStackLayerKernel * * @note Supported input tensor rank: up to 4 diff --git a/arm_compute/runtime/CL/functions/CLStridedSlice.h b/arm_compute/runtime/CL/functions/CLStridedSlice.h index 6fab0c0186..b1edc2481c 100644 --- a/arm_compute/runtime/CL/functions/CLStridedSlice.h +++ b/arm_compute/runtime/CL/functions/CLStridedSlice.h @@ -74,9 +74,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - void configure(const ICLTensor *input, ICLTensor *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + void configure(const ICLTensor *input, + ICLTensor *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); /** Configure kernel * * @note Supported tensor rank: up to 4 @@ -92,9 +97,15 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); /** Static function to check if given info will lead to a valid configuration of @ref CLStridedSlice * @@ -110,9 +121,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); // Inherited methods overridden: void run() override; @@ -143,9 +159,15 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + void configure(const CLCompileContext &compile_context, + const ITensorInfo *input, + ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); /** Static function to check if given info will lead to a valid configuration of @ref CLStridedSlice * @@ -161,9 +183,14 @@ public: * @param[in] shrink_axis_mask (Optional) If the ith bit of shrink_axis_mask is set, it implies that the ith specification shrinks the dimensionality by 1. * A slice of size 1 starting from starts[i] in the dimension must be preserved. */ - static Status validate(const ITensorInfo *input, const ITensorInfo *output, - const Coordinates &starts, const Coordinates &ends, const BiStrides &strides, - int32_t begin_mask = 0, int32_t end_mask = 0, int32_t shrink_axis_mask = 0); + static Status validate(const ITensorInfo *input, + const ITensorInfo *output, + const Coordinates &starts, + const Coordinates &ends, + const BiStrides &strides, + int32_t begin_mask = 0, + int32_t end_mask = 0, + int32_t shrink_axis_mask = 0); }; } // namespace experimental } // namespace arm_compute diff --git a/arm_compute/runtime/CL/functions/CLTile.h b/arm_compute/runtime/CL/functions/CLTile.h index c266adbbd4..4c414670a5 100644 --- a/arm_compute/runtime/CL/functions/CLTile.h +++ b/arm_compute/runtime/CL/functions/CLTile.h @@ -59,7 +59,10 @@ public: * @param[in] multiples Contains the number of times the input tensor should be replicated on the given dimension. * @param[out] output Destination tensor. Same as @p input */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, ICLTensor *output, const Multiples &multiples); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + ICLTensor *output, + const Multiples &multiples); /** Static function to check if given info will lead to a valid configuration of @ref CLTile * * @param[in] input Source tensor info. Data type supported: All. diff --git a/arm_compute/runtime/CL/functions/CLTranspose.h b/arm_compute/runtime/CL/functions/CLTranspose.h index a866aeabaa..9dc977fbeb 100644 --- a/arm_compute/runtime/CL/functions/CLTranspose.h +++ b/arm_compute/runtime/CL/functions/CLTranspose.h @@ -88,6 +88,6 @@ private: struct Impl; std::unique_ptr<Impl> _impl; }; -} +} // namespace arm_compute #endif /* ARM_COMPUTE_CLTRANSPOSE_H */ diff --git a/arm_compute/runtime/CL/functions/CLUnstack.h b/arm_compute/runtime/CL/functions/CLUnstack.h index 32ad439b70..a6eee43177 100644 --- a/arm_compute/runtime/CL/functions/CLUnstack.h +++ b/arm_compute/runtime/CL/functions/CLUnstack.h @@ -26,9 +26,8 @@ #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/core/Types.h" -#include "arm_compute/runtime/IFunction.h" - #include "arm_compute/runtime/CL/functions/CLStridedSlice.h" +#include "arm_compute/runtime/IFunction.h" #include <memory> @@ -72,7 +71,10 @@ public: * @param[in] axis The axis to unstack along. Valid values are [-R,R) where R is the input's rank. Negative values wrap around. * */ - void configure(const CLCompileContext &compile_context, const ICLTensor *input, const std::vector<ICLTensor *> &output_vector, int axis); + void configure(const CLCompileContext &compile_context, + const ICLTensor *input, + const std::vector<ICLTensor *> &output_vector, + int axis); /** Static function to check if given info will lead to a valid configuration of @ref CLUnstack * * @param[in] input Input tensor info. Data type supported: All. diff --git a/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h index adf5f18626..efea9a1550 100644 --- a/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h @@ -84,8 +84,13 @@ public: * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation * available which may introduce a drop of accuracy as well. Default is false */ - void configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + void configure(ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); /** Set the input and output tensors. * * @note: This function only works with 3x3,3x1,1x3,5x5,5x1,1x5,7x1 and 1x7 kernels along with unit strides for both NCHW and NHWC data layout @@ -104,8 +109,14 @@ public: * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation * available which may introduce a drop of accuracy as well. Default is false */ - void configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + void configure(const CLCompileContext &compile_context, + ICLTensor *input, + const ICLTensor *weights, + const ICLTensor *biases, + ICLTensor *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); /** Static function to check if given info will lead to a valid configuration of @ref CLWinogradConvolutionLayer * * @note: This function only works with 3x3,3x1,1x3,5x5,5x1 and 1x5 kernels along with unit strides for both NCHW and NHWC data layout @@ -125,8 +136,13 @@ public: * * @return a status */ - static Status validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info, - const ActivationLayerInfo &act_info = ActivationLayerInfo(), bool enable_fast_math = false); + static Status validate(const ITensorInfo *input, + const ITensorInfo *weights, + const ITensorInfo *biases, + const ITensorInfo *output, + const PadStrideInfo &conv_info, + const ActivationLayerInfo &act_info = ActivationLayerInfo(), + bool enable_fast_math = false); // Inherited methods overridden: void run() override; diff --git a/arm_compute/runtime/CL/tuners/CLTuningParametersList.h b/arm_compute/runtime/CL/tuners/CLTuningParametersList.h index 69572c98d2..5f6d12b4a7 100644 --- a/arm_compute/runtime/CL/tuners/CLTuningParametersList.h +++ b/arm_compute/runtime/CL/tuners/CLTuningParametersList.h @@ -29,6 +29,7 @@ #include "arm_compute/core/Helpers.h" #include "arm_compute/runtime/CL/CLTunerTypes.h" #include "arm_compute/runtime/CL/CLTuningParams.h" + #include "support/ToolchainSupport.h" #include <memory> |