ArmNN
 24.02
Public Attributes | List of all members
GpuFsaPreCompiledBlob Struct Reference

A structure which contains all the elements needed to execute a fused workload in the GpuFsa Backend. More...

#include <GpuFsaBackend.hpp>

Public Attributes

std::unique_ptr< arm_compute::experimental::dynamic_fusion::GpuWorkloadSketch > sketch = nullptr
 
std::shared_ptr< arm_compute::experimental::dynamic_fusion::GpuWorkloadContext > workloadContext = nullptr
 
std::unique_ptr< std::vector< arm_compute::ITensorInfo * > > inputTensorInfos = nullptr
 
std::unique_ptr< std::vector< arm_compute::ITensorInfo * > > outputTensorInfos = nullptr
 

Detailed Description

A structure which contains all the elements needed to execute a fused workload in the GpuFsa Backend.

Parameters
[in,out]sketchA unique pointer to the sketch containing the operators which have been fused.
[in,out]TensorInfosA shared pointer to a GpuWorkloadContext which creates + stores TensorInfos
[in,out]inputTensorInfosA unique pointer to a vector of inputTensorInfos used by the sketch
[in,out]outputTensorInfosA unique pointer to a vector of outputTensorInfos used by the sketch

Definition at line 32 of file GpuFsaBackend.hpp.

Member Data Documentation

◆ inputTensorInfos

std::unique_ptr<std::vector<arm_compute::ITensorInfo*> > inputTensorInfos = nullptr

Definition at line 37 of file GpuFsaBackend.hpp.

Referenced by GpuFsaPreCompiledWorkload::Execute().

◆ outputTensorInfos

std::unique_ptr<std::vector<arm_compute::ITensorInfo*> > outputTensorInfos = nullptr

Definition at line 38 of file GpuFsaBackend.hpp.

Referenced by GpuFsaPreCompiledWorkload::Execute().

◆ sketch

std::unique_ptr<arm_compute::experimental::dynamic_fusion::GpuWorkloadSketch> sketch = nullptr

Definition at line 34 of file GpuFsaBackend.hpp.

Referenced by GpuFsaPreCompiledWorkload::Execute(), armnn::GpuFsaActivationCreateOp(), armnn::GpuFsaBatchMatMulCreateOp(), armnn::GpuFsaCastCreateOp(), armnn::GpuFsaConvolution2dCreateOp(), armnn::GpuFsaDepthwiseConvolution2dCreateOp(), armnn::GpuFsaElementwiseBinaryCreateOp(), armnn::GpuFsaPooling2dCreateOp(), armnn::GpuFsaReshapeCreateOp(), armnn::GpuFsaResizeCreateOp(), armnn::GpuFsaSoftmaxCreateOp(), and GpuFsaBackend::OptimizeSubgraphView().

◆ workloadContext

std::shared_ptr<arm_compute::experimental::dynamic_fusion::GpuWorkloadContext> workloadContext = nullptr

Definition at line 35 of file GpuFsaBackend.hpp.

Referenced by armnn::GpuFsaActivationCreateOp(), armnn::GpuFsaBatchMatMulCreateOp(), armnn::GpuFsaCastCreateOp(), armnn::GpuFsaConvolution2dCreateOp(), armnn::GpuFsaDepthwiseConvolution2dCreateOp(), armnn::GpuFsaElementwiseBinaryCreateOp(), armnn::GpuFsaPooling2dCreateOp(), armnn::GpuFsaReshapeCreateOp(), armnn::GpuFsaResizeCreateOp(), armnn::GpuFsaSoftmaxCreateOp(), and GpuFsaBackend::OptimizeSubgraphView().

The documentation for this struct was generated from the following file: