24.02.1
|
Go to the documentation of this file.
24 #ifndef ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
25 #define ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
42 class ClConvertFullyConnectedWeights;
45 class ClGemmLowpMatrixMultiplyCore;
50 class ClMatMulNativeKernel;
51 class ClMatMulLowpNativeKernel;
139 TransposedWeights = 10,
140 ConvertedWeights = 11,
145 std::unique_ptr<ClConvertFullyConnectedWeights> _convert_weights;
146 std::unique_ptr<ClFlatten> _flatten;
147 std::unique_ptr<ClTranspose> _reshape_weights;
148 std::unique_ptr<ClGemm> _mm_gemm;
149 std::unique_ptr<ClGemmLowpMatrixMultiplyCore> _mm_gemmlowp;
151 std::unique_ptr<kernels::ClMatMulNativeKernel> _matmul_native_kernel;
152 std::unique_ptr<kernels::ClMatMulLowpNativeKernel> _matmul_lowp_native_kernel;
163 bool _run_convert_weights{
false};
164 bool _transpose_weights{
false};
165 bool _dynamic_gemm{
false};
166 bool _use_matmul{
false};
168 bool _is_fc_after_conv{
true};
169 bool _is_quantized{
false};
170 bool _is_prepared{
false};
172 #ifdef ARM_COMPUTE_ASSERTS_ENABLED
173 int _asrt_run_count{};
174 int _asrt_prepare_count{};
175 #endif // ARM_COMPUTE_ASSERTS_ENABLED
179 #endif // ACL_SRC_GPU_CL_OPERATORS_CLFULLYCONNECTED_H
void configure(const CLCompileContext &compile_context, ITensorInfo *src, ITensorInfo *weights, ITensorInfo *biases, ITensorInfo *dst, FullyConnectedLayerInfo fc_info=FullyConnectedLayerInfo())
Set the input and output tensors.
std::vector< MemoryInfo > MemoryRequirements
SimpleTensor< float > src
Fully connected layer info.
Basic interface for functions which have a single async CL kernel.
void run(ITensorPack &tensors) override
Run the kernels contained in the function.
Basic function to compute a Fully Connected layer on OpenCL.
experimental::MemoryRequirements workspace() const override
Return the memory requirements required by the workspace.
Store the tensor's metadata.
Copyright (c) 2017-2024 Arm Limited.
void prepare(ITensorPack &tensors) override
Prepare the function for executing.
Store the tensor's metadata.
static Status validate(const ITensorInfo *src, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *dst, FullyConnectedLayerInfo fc_info=FullyConnectedLayerInfo())
Static function to check if given info will lead to a valid configuration.