write_aop_knowledge — Write knowledge about hardware dependent behavior of automatic operator
parallelization to file.
write_aop_knowledge( : : FileName, GenParamName, GenParamValue : )
write_aop_knowledge stores the automatic parallelization knowledge
of HALCON operators to file. The knowledge hold by the HALCON system is
used to better utilize multiprocessor
hardware in order to speed up the processing of operators. To optimize the
automatic parallelization of operators, HALCON needs some specific
knowledge about the speed of operators on the used hardware.
This hardware-specific knowledge
can be obtained by calling the operator optimize_aop or read from
file by using read_aop_knowledge, respectively.
With write_aop_knowledge it is possible to store this knowledge
explicitly in a binary file. At this, FileName denotes the
name of this file (incl. path and file extension). If an empty string
'' is passed to FileName, HALCON writes the knowledge
to the specific file '.aop_info' in the HALCON installation
directory (Linux/macOS) or within the common application data folder
(Windows). This file is read by HALCON during the initialization phase
automatically.
write_aop_knowledge adds attributes to the knowledge like
the host name and information about HALCON architecture, variant,
version, and revision. This enables HALCON to write aop knowledge
sets of different machines or HALCON systems into the same file.
Therefore, write_aop_knowledge can control the updating of the
files contents by a pair of values passed to GenParamName and
GenParamValue. Every entry in
GenParamName must have one corresponding, specifying entry in
GenParamValue, meaning that the tuples passed to the parameters
must have the same length. GenParamName supports the values in
following list, describing the possible applicable values for
GenParamValue:
sets the way how existing knowledge with same attributes would be updated in the file.
for GenParamValue deletes
the existing knowledge with matching attributes before the new
information is added to file.
overwrites existing knowledge and adds new one (default).
keeps all existing operator information and just adds the knowledge not already contained.
however refuses any update of the file but checks the user priviledges, i.e., if read write access is available.
denotes that the corresponding index value of
GenParamValue contains an operator name. It delimits the storage
on a dedicated operator set. Multiple definitions
of operators are possible. By default, information of any
operator possessing AOP optimization data is written.
focuses on AOP information for a specific
iconic type, specified by following values of GenParamValue:
'byte','uint2','real','int1',
'int2','int4','int8','direction',
'vector_field','cyclic',ValRef{'vector_field},
'complex','region','xld'. Multiple definitions
of iconic types are possible.
FileName (input_control) filename.write → (string)
Name of knowledge file
Default value: ''
GenParamName (input_control) string(-array) → (string)
Parameter name
Default value: 'none'
Suggested values: 'mode', 'operator', 'iconic_type'
GenParamValue (input_control) string(-array) → (string / integer / real)
Parameter value
Number of elements: GenParamName == GenParamValue
Default value: 'none'
Suggested values: 'truncate', 'renew', 'append'
write_aop_knowledge returns 2 (H_MSG_TRUE) if all parameters are correct and
the file could be read. If necessary, an exception is raised.
Foundation