get_dl_device_paramT_get_dl_device_paramGetDlDeviceParamGetDlDeviceParamget_dl_device_param (Operator)

Name

get_dl_device_paramT_get_dl_device_paramGetDlDeviceParamGetDlDeviceParamget_dl_device_param — Return the parameters of a deep-learning-capable hardware device.

Signature

Description

get_dl_device_paramget_dl_device_paramGetDlDeviceParamGetDlDeviceParamget_dl_device_param returns the parameter values GenParamValueGenParamValueGenParamValuegenParamValuegen_param_value of GenParamNameGenParamNameGenParamNamegenParamNamegen_param_name for the deep-learning-capable hardware device (hereafter referred to as device) DLDeviceHandleDLDeviceHandleDLDeviceHandleDLDeviceHandledldevice_handle. See query_available_dl_devicesquery_available_dl_devicesQueryAvailableDlDevicesQueryAvailableDlDevicesquery_available_dl_devices for details about deep-learning-capable hardware devices.

Supported values for GenParamNameGenParamNameGenParamNamegenParamNamegen_param_name are:

'calibration_precisions'"calibration_precisions""calibration_precisions""calibration_precisions""calibration_precisions":

Specifies the unit data types that can be used for a calibration of a deep learning model.

Possible values: 'int8'"int8""int8""int8""int8".

'cast_precisions'"cast_precisions""cast_precisions""cast_precisions""cast_precisions":

Specifies the unit data types that can be used for a cast of a deep learning model.

When changing the data type the calibration does not require any images.

Possible values: 'float32'"float32""float32""float32""float32", 'float16'"float16""float16""float16""float16".

'conversion_supported'"conversion_supported""conversion_supported""conversion_supported""conversion_supported":

Returns 'true'"true""true""true""true" if unit data types for either a calibration or a cast of a deep learning model are available. Returns 'false'"false""false""false""false" in any other case.

'id'"id""id""id""id":

The ID of the device. Within each inference engine, the IDs of its supported devices are unique. The same holds for devices supported through HALCON.

'inference_only'"inference_only""inference_only""inference_only""inference_only":

Indicates if the device can only be used to infer deep learning models ('true'"true""true""true""true") or also supports training or gradient-based operations ('false'"false""false""false""false").

'ai_accelerator_interface'"ai_accelerator_interface""ai_accelerator_interface""ai_accelerator_interface""ai_accelerator_interface":

AI Accelerator Interface (AI ²) on which this unit DLDeviceHandleDLDeviceHandleDLDeviceHandleDLDeviceHandledldevice_handle is executed. In case the device is directly supported by HALCON, the value 'none'"none""none""none""none" is returned.

Possible values: 'tensorrt'"tensorrt""tensorrt""tensorrt""tensorrt", 'openvino'"openvino""openvino""openvino""openvino", 'none'"none""none""none""none".

'info'"info""info""info""info":

Dictionary containing additional information on the device.

Restriction: Only for devices that are supported via an AI2-interface.

'name'"name""name""name""name":

Name of the device.

'optimize_for_inference_params'"optimize_for_inference_params""optimize_for_inference_params""optimize_for_inference_params""optimize_for_inference_params":

Dictionary with default-defined conversion parameters for a calibration or cast operation of a deep learning model. The entries can be changed.

In case no parameter applies to the set device, an empty dictionary is returned.

Restriction: Only for devices that are supported via an AI2-interface.

'precisions'"precisions""precisions""precisions""precisions":

Specifies the data types that the unit supports for the weights and/or activations of a deep-learning-based model.

Possible values: 'float32'"float32""float32""float32""float32", 'float16'"float16""float16""float16""float16", 'int8'"int8""int8""int8""int8".

'settable_device_params'"settable_device_params""settable_device_params""settable_device_params""settable_device_params":

Dictionary with settable device parameters.

Restriction: Only for devices that are supported via an AI2-interface.

'type'"type""type""type""type":

Type of the device.

Execution Information

• Multithreading type: reentrant (runs in parallel with non-exclusive operators).
• Multithreading scope: global (may be called from any thread).
• Processed without parallelization.

Parameters

Result

If the parameters are valid, the operator get_dl_device_paramget_dl_device_paramGetDlDeviceParamGetDlDeviceParamget_dl_device_param returns the value 2 ( H_MSG_TRUE) . If necessary, an exception is raised.

Possible Predecessors

query_available_dl_devicesquery_available_dl_devicesQueryAvailableDlDevicesQueryAvailableDlDevicesquery_available_dl_devices

Possible Successors

set_dl_model_paramset_dl_model_paramSetDlModelParamSetDlModelParamset_dl_model_param

Module

Foundation