- Generated by
1.9.1
|
NVIDIA NvNeural SDK
2022.2
GPU inference framework for NVIDIA Nsight Deep Learning Designer
|
The DimensionParamTypeDesc further describes a parameter for the plugin and is pointed to from the higher level parameter descriptor ParamDesc. More...
#include <nvneural/CoreTypes.h>
Public Attributes | |
| BaseParamTypeDesc | baseDesc |
| This is the base param desc struct which all xParamTypeDesc must have. | |
| std::size_t | maxDimensions |
| This is the maximum amount of dimensions that are valid. | |
| std::size_t | minDimensions |
| This is the minimum amount of dimensions that are needed. | |
| std::size_t | paramDependenciesCount |
| This is the amount of parameter dependencies contained in the memory block. | |
| const ParamDependencyDesc * | pParamDependencies |
| This is the array of parameter dependency with paramDependenciesCount members. More... | |
The DimensionParamTypeDesc further describes a parameter for the plugin and is pointed to from the higher level parameter descriptor ParamDesc.
In this case, it describes a dimension, using a max and min dimension.
An example would be needing height and width, in HxW format. In this case, the maxDimensions and the minDimensions would be set to 2.
Another example, would be needing height and width, but optionally channel. This could be in the format CxHxW or HxW. In this case, the maxDimensions would be set to 3 and the minDimensions would be set to 2.
| const ParamDependencyDesc* nvneural::DimensionParamTypeDesc::pParamDependencies |
This is the array of parameter dependency with paramDependenciesCount members.
Each dependency is treated as an AND in the final evaluation. If the same parameter name appears multiple time with different values then all its listed parameter value will be treated as an OR.
Examples: 1) pParamDependencies = [{"paramA", "valueA"}, {"paramB", "valueB"}] result in the following expression => (paramA with valueA) && (paramB with valueB) 2) pParamDependencies = [{"paramA", "valueA"}, {"paramA", "valueOpt"}, {"paramB", "valueB"}] result in the following expression => (paramA with valueA || paramA with valueOpt) && (paramB with valueB)
1.9.1