![]() |
MMALIB User Guide
|
Structure containing the parameters for input to the execute phase of CNN convolution computation These parameters will not exist in J7AM, kept for J7ES compatibility.
Definition at line 194 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
Data Fields | |
int32_t | validColsIn |
Valid columns in a row of input feature maps for one call of processing of the kernel for non strided convolution and for 1x1 stride 2 for execute phase More... | |
int32_t | validColsPerRowIn |
Valid columns in a row of input feature maps for one call of processing of the kernel for strided convolution except for 1x1 stride 2 for execute phase More... | |
int32_t | validRowsIn |
Valid input rows of input feature maps for one call of processing of the kernel for strided convolution except for 1x1 stride 2 for execute phase More... | |
int32_t | col |
column offset of each row of input feature map enables with a flag in execute phase More... | |
int32_t | subMChannels |
Number of output channels per kernel call in execute phase More... | |
int32_t | quantMethod |
quantMethod 0 - per call 1 - per channel More... | |
int32_t | padFillValue |
pad fill value More... | |
MMALIB_enableDynamicRange_type | enableDynamicRange |
Flag indicating whether to enable dynamic range range (min,max) calculation. More... | |
MMALIB_initDynamicRange_type | initDynamicRange |
Flag indicating whether to reset dynamic range (min,max) calculation at beginning of compute. More... | |
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::validColsIn |
Valid columns in a row of input feature maps for one call of processing of the kernel for non strided convolution and for 1x1 stride 2 for execute phase
Definition at line 199 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::validColsPerRowIn |
Valid columns in a row of input feature maps for one call of processing of the kernel for strided convolution except for 1x1 stride 2 for execute phase
Definition at line 203 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::validRowsIn |
Valid input rows of input feature maps for one call of processing of the kernel for strided convolution except for 1x1 stride 2 for execute phase
Definition at line 207 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::col |
column offset of each row of input feature map enables with a flag in execute phase
Definition at line 210 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::subMChannels |
Number of output channels per kernel call in execute phase
Definition at line 212 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::quantMethod |
quantMethod 0 - per call 1 - per channel
Definition at line 214 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
int32_t MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::padFillValue |
pad fill value
Definition at line 216 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
MMALIB_enableDynamicRange_type MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::enableDynamicRange |
Flag indicating whether to enable dynamic range range (min,max) calculation.
Definition at line 218 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.
MMALIB_initDynamicRange_type MMALIB_CNN_convolveBias_row_ixX_ixX_oxX_ExecInArgs::initDynamicRange |
Flag indicating whether to reset dynamic range (min,max) calculation at beginning of compute.
Definition at line 220 of file MMALIB_CNN_convolveBias_row_ixX_ixX_oxX.h.