Detailed Description

This structure define the parameters Convolution Layer in TIDL.

Data Fields
int32_t	weights

int32_t	bias

int32_t	perChannelWeightScaleOffset

int32_t	convolutionType

int32_t	numInChannels

int32_t	numOutChannels

int32_t	numGroups

int32_t	kernelW

int32_t	kernelH

int32_t	strideW

int32_t	strideH

int32_t	dilationW

int32_t	dilationH

int32_t	padW

int32_t	padH

float32_tidl	weightScale

float32_tidl	biasScale

int32_t	weightsQ

int32_t	zeroWeightValue

int32_t	biasB

int32_t	biasQ

int32_t	inDataQ

int32_t	outDataQ

int32_t	interDataQ

int32_t	enableBias

int32_t	enablePooling

int32_t	enableEltWise

int32_t	enableEWRelU

int32_t	kernelType

int32_t	enableDepthToSpace

int32_t	upscaleFactor

sTIDL_PoolingParams_t	poolParams

Field Documentation

◆ weights

int32_t sTIDL_ConvParams_t::weights

Offset to the kernel parameters

◆ bias

int32_t sTIDL_ConvParams_t::bias

Offset to the bias parameters

◆ perChannelWeightScaleOffset

int32_t sTIDL_ConvParams_t::perChannelWeightScaleOffset

Offset where per channel weight scales are stored for depthwise convolution layers. A value of zero indicates that this infomation is not valid

◆ convolutionType

int32_t sTIDL_ConvParams_t::convolutionType

Type of the convolution, Reserved for future use

◆ numInChannels

int32_t sTIDL_ConvParams_t::numInChannels

Number of input channels channels to be processed

◆ numOutChannels

int32_t sTIDL_ConvParams_t::numOutChannels

Number of output channels to be processed

◆ numGroups

int32_t sTIDL_ConvParams_t::numGroups

Number of groups in the convolutions

◆ kernelW

int32_t sTIDL_ConvParams_t::kernelW

Kernel width

◆ kernelH

int32_t sTIDL_ConvParams_t::kernelH

Kernel height

◆ strideW

int32_t sTIDL_ConvParams_t::strideW

Stride in horizontal direction

◆ strideH

int32_t sTIDL_ConvParams_t::strideH

Stride in vertical direction

◆ dilationW

int32_t sTIDL_ConvParams_t::dilationW

Dilation in horizontal direction

◆ dilationH

int32_t sTIDL_ConvParams_t::dilationH

Dilation in vertical direction

◆ padW

int32_t sTIDL_ConvParams_t::padW

Horizontal Padding requirement in number of elements

◆ padH

int32_t sTIDL_ConvParams_t::padH

Vertical Padding requirement in number of elements

◆ weightScale

float32_tidl sTIDL_ConvParams_t::weightScale

Floating point scale on Kernel weights

◆ biasScale

float32_tidl sTIDL_ConvParams_t::biasScale

Floating point scale on bias

◆ weightsQ

int32_t sTIDL_ConvParams_t::weightsQ

Q value of Kernel weights

◆ zeroWeightValue

int32_t sTIDL_ConvParams_t::zeroWeightValue

value of weights added for dynamic quantSytle

◆ biasB

int32_t sTIDL_ConvParams_t::biasB

Not used

◆ biasQ

int32_t sTIDL_ConvParams_t::biasQ

Q value kernel Bias

◆ inDataQ

int32_t sTIDL_ConvParams_t::inDataQ

Q value expected for in data

◆ outDataQ

int32_t sTIDL_ConvParams_t::outDataQ

Q value expected for out data

◆ interDataQ

int32_t sTIDL_ConvParams_t::interDataQ

Q value intermediate output data

◆ enableBias

int32_t sTIDL_ConvParams_t::enableBias

Enable/Disable output bias

◆ enablePooling

int32_t sTIDL_ConvParams_t::enablePooling

Enable/Disable 2x2 Spatial pooling

◆ enableEltWise

int32_t sTIDL_ConvParams_t::enableEltWise

Enable/Disable EltWise

◆ enableEWRelU

int32_t sTIDL_ConvParams_t::enableEWRelU

Enable/Disable Relu for EltWise

◆ kernelType

int32_t sTIDL_ConvParams_t::kernelType

Defines the different types of optimizations in kernel types supported by TIDL

◆ enableDepthToSpace

int32_t sTIDL_ConvParams_t::enableDepthToSpace

Enable/Disable depth to Space layer

◆ upscaleFactor

int32_t sTIDL_ConvParams_t::upscaleFactor

Used only when depth to Space layer is enabled - This is derived from numOutChannels and data Params Out num channels

◆ poolParams

sTIDL_PoolingParams_t sTIDL_ConvParams_t::poolParams

Used only if enablePooling is true