get_aop_infoT_get_aop_infoGetAopInfoGetAopInfoget_aop_info (算子)

名称

get_aop_infoT_get_aop_infoGetAopInfoGetAopInfoget_aop_info — 为算子返回 AOP（Automatic Operator Parallelization，自动算子并行化）信息。

签名

描述

HALCON supports a mechanism to optimize the behavior of automatic operator parallelization (AOP) for a given hardware of a machine. The information for this optimization can be determined by the operator optimize_aopoptimize_aopOptimizeAopOptimizeAopOptimizeAopoptimize_aop and can also be partly modified by the operator set_aop_infoset_aop_infoSetAopInfoSetAopInfoSetAopInfoset_aop_info。

To address specific AOP information, the operator's optimization data is indexed within a 3-ary hierarchy that can be obtained by query_aop_infoquery_aop_infoQueryAopInfoQueryAopInfoQueryAopInfoquery_aop_info。The three hierarchy stages are indexed by the operator name, passed in OperatorNameOperatorNameOperatorNameOperatorNameoperatorNameoperator_name, the iconic type, and a parameter string value denoting a special mode or method supported by the indexed operator. The latter two indices are passed by IndexNameIndexNameIndexNameIndexNameindexNameindex_name and IndexValueIndexValueIndexValueIndexValueindexValueindex_value. IndexNameIndexNameIndexNameIndexNameindexNameindex_name holds the dimension identifier whereas IndexValueIndexValueIndexValueIndexValueindexValueindex_value holds the index value as returned in query_aop_infoquery_aop_infoQueryAopInfoQueryAopInfoQueryAopInfoquery_aop_info。

get_aop_infoget_aop_infoGetAopInfoGetAopInfoGetAopInfoget_aop_info returns the specific information parts of an operator's AOP knowledge in InfoValueInfoValueInfoValueInfoValueinfoValueinfo_value if exactly one index per stage was set. The scope of information is specified by InfoNameInfoNameInfoNameInfoNameinfoNameinfo_name and supports following values:

'max_threads'"max_threads""max_threads""max_threads""max_threads""max_threads"

returns the maximum allowed thread number. If no optimization data is stored, InfoValueInfoValueInfoValueInfoValueinfoValueinfo_value contains -1. In case the specified operator does not support automatic parallelization, 1 is returned.

'split_level'"split_level""split_level""split_level""split_level""split_level"

returns all allowed data split levels of the automatic parallelization for this operator, the specified iconic type and parameter value. Possible levels are 'split_tuple'"split_tuple""split_tuple""split_tuple""split_tuple""split_tuple", 'split_channel'"split_channel""split_channel""split_channel""split_channel""split_channel", 'split_domain'"split_domain""split_domain""split_domain""split_domain""split_domain", and 'split_partial'"split_partial""split_partial""split_partial""split_partial""split_partial". If no split level is supported or every level was switched off for the indexed operator an empty string ''"""""""""" is returned. Use get_operator_infoget_operator_infoGetOperatorInfoGetOperatorInfoGetOperatorInfoget_operator_info to query all split levels generally supported by an operator.

'model'"model""model""model""model""model"

returns the stored model type. Possible values are 'threshold'"threshold""threshold""threshold""threshold""threshold", 'linear'"linear""linear""linear""linear""linear", and 'mlp'"mlp""mlp""mlp""mlp""mlp" (see also the description of operator optimize_aopoptimize_aopOptimizeAopOptimizeAopOptimizeAopoptimize_aop). If no optimization data is stored, an empty string is returned.

执行信息

• 多线程类型：独占（仅与独立算子并行运行）。
• 多线程作用域：局部（仅可在创建窗口、模型或工具实例的同一线程中调用）。
• 未采用并行化处理。

参数

结果

get_aop_infoget_aop_infoGetAopInfoGetAopInfoGetAopInfoget_aop_info 在所有参数正确时返回 2 ( H_MSG_TRUE )。如有必要，则抛出异常。

模块

基础