Compute Library
 22.05
arm_compute::opencl Namespace Reference

Interface to convert the 2D Fully Connected weights from NCHW to NHWC or vice versa. More...

Namespaces

 kernels
 

Data Structures

class  ClAbs
 Basic function to get the absolute value of an src tensor. More...
 
class  ClActivation
 Basic function to run kernels::ClActivationKernel. More...
 
class  ClAdd
 Basic function to run arithmetic addition. More...
 
class  CLAuxTensorHandler
 
class  ClCast
 Basic function to run kernels::ClCastKernel. More...
 
class  ClComplexMul
 Basic function to run opencl::kernels::ClComplexMulKernel. More...
 
class  ClConcatenate
 Basic function to execute concatenate tensors along a given axis. More...
 
class  ClConv2d
 Basic function to compute the convolution layer. More...
 
class  ClConvertFullyConnectedWeights
 Basic function to run kernels::ClConvertFullyConnectedWeightsKernel. More...
 
class  ClCopy
 Basic function to run kernels::ClCopyKernel. More...
 
class  ClCrop
 Basic function to run kernels::ClCropKernel. More...
 
class  ClDequantize
 Basic function to run kernels::ClDequantizeKernel that dequantizes an input tensor. More...
 
class  ClDirectConv2d
 Basic function to simulate a directly convolution layer. More...
 
class  ClDirectConv3d
 Basic function to simulate a directly convolution layer with 3 spatial dimensions. More...
 
class  ClElementwiseDivision
 Basic function to run opencl::kernels::ClArithmeticKernel for division. More...
 
class  ClElementwiseMax
 Basic function to run opencl::kernels::ClArithmeticKernel for max. More...
 
class  ClElementwiseMin
 Basic function to run opencl::kernels::ClArithmeticKernel for min. More...
 
class  ClElementwisePower
 Basic function to run opencl::kernels::ClArithmeticKernel for power. More...
 
class  ClElementwiseSquaredDiff
 Basic function to run opencl::kernels::ClArithmeticKernel for squared difference. More...
 
class  ClExp
 Basic function to perform exponential on an src tensor. More...
 
class  ClFill
 Basic function to run kernels::ClFillKernel. More...
 
class  ClFlatten
 Basic function to flatten a given input. More...
 
class  ClFloor
 Basic function to run kernels::ClFloorKernel. More...
 
class  ClFullyConnected
 Basic function to compute a Fully Connected layer on OpenCL. More...
 
class  ClGemm
 Basic function to execute GEMM on OpenCL. More...
 
class  ClGemmConv2d
 Basic function to compute the convolution layer. More...
 
class  ClGemmLowpMatrixMultiplyCore
 Basic function to execute GEMMLowpMatrixMultiplyCore on OpenCL. More...
 
class  ClGemmLowpOutputStage
 Basic function to execute GEMMLowpQuantizeDown kernels on CL. More...
 
class  ClKernelLibrary
 ClKernelLibrary contains all the OpenCL kernels that are used throughout the library. More...
 
class  ClLog
 Basic function to perform elementwise log on an src tensor. More...
 
class  ClLogicalNot
 Basic function to run kernels::ClElementWiseUnaryKernel for NOT operation. More...
 
class  ClMul
 Basic function to run opencl::kernels::ClMulKernel. More...
 
class  ClNeg
 Basic function to negate an src tensor. More...
 
class  ClPermute
 Basic function to run kernels::ClPermuteKernel. More...
 
class  ClPool2d
 Basic function to simulate a pooling layer with the specified pooling operation. More...
 
class  ClPool3d
 Basic function to simulate a pooling layer with the specified pooling operation. More...
 
class  ClPRelu
 Basic operator to run arm_compute::opencl::kernels::ClArithmeticKernel for PRELU. More...
 
class  ClQuantize
 Basic function to run kernels::ClQuantizeKernel that dequantizes an input tensor. More...
 
class  ClReshape
 Basic function to run kernels::ClReshapeKernel. More...
 
class  ClRound
 Basic function to get the round (to the nearest even) value of an src tensor. More...
 
class  ClRsqrt
 Basic function to perform inverse square root on an src tensor. More...
 
class  ClScale
 Basic function to simulate a scale layer. More...
 
class  ClSin
 Basic function to calculate sine of an src tensor. More...
 
class  ClSoftmax
 
class  ClSub
 Basic function to run arithmetic subtraction. More...
 
class  ClTranspose
 Basic function to run kernels::ClTransposeKernel. More...
 
class  ClWinogradConv2d
 Basic function to execute Winograd-based convolution on OpenCL. More...
 

Typedefs

using ClCompileContext = arm_compute::CLCompileContext
 
using IClKernel = arm_compute::ICLKernel
 
using IClOperator = experimental::ICLOperator
 
using KernelType = kernels::ClArithmeticKernel
 

Detailed Description

Interface to convert the 2D Fully Connected weights from NCHW to NHWC or vice versa.

Note
This function can be applied to the 2D weights used by a Fully Connected layer if:
  • It follows a Convolution layer
  • The data layout used by the network does not match the one the model has been trained in.
This function assumes the weights are already reshaped (transposed)

Typedef Documentation

◆ ClCompileContext

◆ IClKernel

Definition at line 34 of file IClKernel.h.

◆ IClOperator

Definition at line 34 of file IClOperator.h.

◆ KernelType

Definition at line 34 of file ClPRelu.cpp.