![]() |
MMALIB User Guide
|
This structure holds all the input parameters for reordering CNN filter weights for column-based convolution kernel.
Definition at line 85 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
Data Fields | |
int32_t | dataType |
Data type must be a member of MMALIB_data_type_e. More... | |
int32_t | Ni |
Number of input channels to be processed. More... | |
int32_t | No |
Number of output channels to be processed. More... | |
int32_t | Fr |
Kernel height. More... | |
int32_t | Fc |
Kernel width. More... | |
int32_t | strideX |
Stride in the horizontal dimension. More... | |
int32_t | strideY |
Stride in the vertical dimension. More... | |
int32_t | dilationX |
Dilation in the horizontal dimension. More... | |
int32_t | featureWidth |
Total feature width. More... | |
int32_t | blockFeatureHeight |
Number of rows from the input feature map to be processed, including top and bottom pad that is to be applied inside the kernel. More... | |
int32_t | topPad |
Pixels of pad to apply to the top side of the input data. More... | |
int32_t | bottomPad |
Pixels of pad to apply to the bottom side of the input data. More... | |
int32_t | leftPad |
Pixels of pad to apply to the left side of the input data. More... | |
int32_t | rightPad |
Pixels of pad to apply to the rights side of the input data. More... | |
int32_t | numBiasVals |
Number of elements used for the bias (cols in weights, rows in feature maps) More... | |
int32_t | numGroupsPerKernel |
Number of groups to be processed in kernel call. More... | |
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::dataType |
Data type must be a member of MMALIB_data_type_e.
Definition at line 87 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::Ni |
Number of input channels to be processed.
Definition at line 88 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::No |
Number of output channels to be processed.
Definition at line 89 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::Fr |
Kernel height.
Definition at line 90 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::Fc |
Kernel width.
Definition at line 91 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::strideX |
Stride in the horizontal dimension.
Definition at line 92 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::strideY |
Stride in the vertical dimension.
Definition at line 93 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::dilationX |
Dilation in the horizontal dimension.
Definition at line 94 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::featureWidth |
Total feature width.
Definition at line 95 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::blockFeatureHeight |
Number of rows from the input feature map to be processed, including top and bottom pad that is to be applied inside the kernel.
Definition at line 96 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::topPad |
Pixels of pad to apply to the top side of the input data.
Definition at line 97 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::bottomPad |
Pixels of pad to apply to the bottom side of the input data.
Definition at line 98 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::leftPad |
Pixels of pad to apply to the left side of the input data.
Definition at line 99 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::rightPad |
Pixels of pad to apply to the rights side of the input data.
Definition at line 100 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::numBiasVals |
Number of elements used for the bias (cols in weights, rows in feature maps)
Definition at line 101 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.
int32_t MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights_Args::numGroupsPerKernel |
Number of groups to be processed in kernel call.
Definition at line 102 of file MMALIB_CNN_convolve_col_smallNo_ixX_ixX_oxX_reorderWeights.h.