ONE - On-device Neural Engine
|
#include <CustomKernel.h>
Public Member Functions | |
CustomKernel (nnfw_custom_eval evalFunction) | |
virtual void | configure (backend::custom::CustomKernelConfigParams &&inParams) |
void | run () override |
Public Member Functions inherited from onert::exec::IFunction | |
virtual | ~IFunction ()=default |
virtual void | prepare () |
Data Fields | |
backend::custom::CustomKernelConfigParams | _in_params |
char * | _userdata |
size_t | _userdata_size |
nnfw_custom_eval | _evalFunction |
Definition at line 32 of file CustomKernel.h.
|
explicit |
Definition at line 65 of file CustomKernel.cc.
|
virtual |
Fills _params field used later by user specified eval function
inParams | custom kernel parameters |
Definition at line 70 of file CustomKernel.cc.
References _in_params, _userdata, and _userdata_size.
|
overridevirtual |
Implements onert::exec::IFunction.
Definition at line 78 of file CustomKernel.cc.
References _evalFunction, _in_params, _userdata, _userdata_size, onert::api::APIConverter::convertOperand(), onert::backend::custom::CustomKernelConfigParams::input_tensors, onert::backend::custom::CustomKernelConfigParams::input_types, nnfw_custom_kernel_params::inputs, nnfw_custom_kernel_params::ninputs, nnfw_custom_kernel_params::noutputs, onert::backend::custom::CustomKernelConfigParams::output_tensors, onert::backend::custom::CustomKernelConfigParams::output_types, and nnfw_custom_kernel_params::outputs.
Referenced by package.infer.session::inference().
nnfw_custom_eval onert::api::CustomKernel::_evalFunction |
Definition at line 42 of file CustomKernel.h.
Referenced by run().
backend::custom::CustomKernelConfigParams onert::api::CustomKernel::_in_params |
Definition at line 37 of file CustomKernel.h.
Referenced by configure(), and run().
char* onert::api::CustomKernel::_userdata |
Definition at line 39 of file CustomKernel.h.
Referenced by configure(), and run().
size_t onert::api::CustomKernel::_userdata_size |
Definition at line 40 of file CustomKernel.h.
Referenced by configure(), and run().