24.02.1
|
Data Structures | |
class | ClDWCNativeConfigArray |
Basic container for the OpenCL depthwise convolution configuration functions. More... | |
class | ClDWCNativeDefaultConfigBifrost |
Bifrost based OpenCL depthwise convolution configuration. More... | |
class | ClDWCNativeDefaultConfigValhall |
Valhall based OpenCL depthwise convolution configuration. More... | |
class | ClDWCNativeKernelConfigurationFactory |
ClDWCNativeKernelConfigurationFactory factory class. More... | |
class | IClDWCNativeKernelConfig |
Basic interface for the depthwise convolution kernel configuration. More... | |
Functions | |
bool | use_cl_image_for_weights (const ITensorInfo *weights, unsigned int depth_multiplier) |
Utility function to know whether we can use the cl image storage for the weights of depthwise convolution to get better performance. More... | |
bool use_cl_image_for_weights | ( | const ITensorInfo * | weights, |
unsigned int | depth_multiplier | ||
) |
Utility function to know whether we can use the cl image storage for the weights of depthwise convolution to get better performance.
[in] | weights | Weights TensorInfo of the depthwise convolution |
[in] | depth_multiplier | Depth multiplier |
Definition at line 32 of file ClDWCNativeHeuristicsHelpers.cpp.
References ITensorInfo::data_layout(), arm_compute::export_to_cl_image(), arm_compute::get_data_layout_dimension_index(), arm_compute::HEIGHT, ITensorInfo::tensor_shape(), and arm_compute::WIDTH.
Referenced by GpuCkwDirectConv2d::write_component_code().