Optimized Belief Propagation (CPU and GPU)
ParallelParams Class Referenceabstract

Abstract class for holding and processing parallelization parameters.
Child class(es) specific to implementation(s) must be defined. More...

#include <ParallelParams.h>

Inheritance diagram for ParallelParams:

Public Member Functions

virtual void SetParallelDims (const std::array< unsigned int, 2 > &parallelDims)=0
 Set parallel parameters for each kernel to the same input dimensions. More...
 
virtual RunData AsRunData () const =0
 Retrieve current parallel parameters as RunData type. More...
 
virtual void AddTestResultsForParallelParams (const std::array< unsigned int, 2 > &p_params_curr_run, const RunData &curr_run_data)=0
 Add results from run with same specified parallel parameters used every parallel component. More...
 
virtual void SetOptimizedParams ()=0
 Retrieve optimized parameters from results across multiple runs with different parallel parameters and set current parameters to retrieved optimized parameters. More...
 
virtual std::array< unsigned int, 2 > OptParamsForKernel (const std::array< unsigned int, 2 > &kernel_location) const =0
 Get optimized parallel parameters for parallel processing kernel for kernel that is indexed as an array of two unsigned integers. More...
 

Detailed Description

Abstract class for holding and processing parallelization parameters.
Child class(es) specific to implementation(s) must be defined.

Definition at line 39 of file ParallelParams.h.

Member Function Documentation

◆ AddTestResultsForParallelParams()

virtual void ParallelParams::AddTestResultsForParallelParams ( const std::array< unsigned int, 2 > &  p_params_curr_run,
const RunData curr_run_data 
)
pure virtual

Add results from run with same specified parallel parameters used every parallel component.

Parameters
p_params_curr_run
curr_run_data

Implemented in ParallelParamsBp.

◆ AsRunData()

virtual RunData ParallelParams::AsRunData ( ) const
pure virtual

Retrieve current parallel parameters as RunData type.

Returns
Current parallel parameters data

Implemented in ParallelParamsBp.

◆ OptParamsForKernel()

virtual std::array<unsigned int, 2> ParallelParams::OptParamsForKernel ( const std::array< unsigned int, 2 > &  kernel_location) const
pure virtual

Get optimized parallel parameters for parallel processing kernel for kernel that is indexed as an array of two unsigned integers.

Parameters
kernel_location
Returns
Optimized parallel parameters for indexed kernel

Implemented in ParallelParamsBp.

Here is the caller graph for this function:

◆ SetOptimizedParams()

virtual void ParallelParams::SetOptimizedParams ( )
pure virtual

Retrieve optimized parameters from results across multiple runs with different parallel parameters and set current parameters to retrieved optimized parameters.

Implemented in ParallelParamsBp.

◆ SetParallelDims()

virtual void ParallelParams::SetParallelDims ( const std::array< unsigned int, 2 > &  parallelDims)
pure virtual

Set parallel parameters for each kernel to the same input dimensions.

Parameters
parallelDims

Implemented in ParallelParamsBp.


The documentation for this class was generated from the following file: