Software and hardware restrictions

AI Inference Server Function Manual
Portfolio
Industrial AI
Product
AI Inference Server
Software version
2.1.0
Language
en-US
CAUTION:

AI Inference Server should not be used in mission-critical scenarios with high risks (i.e. development, construction, maintenance or operation of systems, the failure of which could lead to a life-threatening situation or to catastrophic damage ("critical application")).

Examples of critical applications: Use in avionics, navigation, autonomous vehicle applications, AI solutions for automotive products, the military, medicine, life support or other life-critical applications.

Note the following software and hardware limitations:

  • Only one AI Inference Server application instance can be installed on an Edge device. If you attempt to install more than one application instance with different MLFB numbers, you receive error messages and cannot use AI Inference Server.

  • AI Inference Server uses ready-to-use connectors of the Siemens Industrial Edge platform. The performance limitations for the connectors are listed in the respective help file of each app and connector.

  • The performance of the connectors and the resolutions of the collected data are the responsibility of each connector provided by the Edge platform. Restrictions and limitations are documented in the respective user manual.

  • AI Inference Server can only browse tags that are available on the data bus of the installed Edge device. The Edge platform in its current version does not permit access to the data bus of other Edge devices.

  • In this version, the tag browsing function only includes tags that are available on the Databus from the SIMATIC S7 Connector.

  • MLFB 6AV2170-0LA10-0AA0 (Single-Pipeline-Version of the AI Inference Server) has been tested on SIMATIC IPC227E, SIMATIC IPC427E, SIMATIC IPC847E, SIMATIC IPC BX-39A and Virtual Edge Device.

  • MLFB 6AV2170-0LA10-1AA0 (Multiple Pipeline Version of the AI Inference Servers) has been tested on SIMATIC IPC427E, SIMATIC IPC847E, SIMATIC IPC BX-39A and Virtual Edge Device.

  • MLFB 6AV2170-0LA11-0AA0 (AI Inference Server - GPU accelerated - 1 Pipeline) has been tested on SIMATIC IPC427E, SIMATIC IPC847E, SIMATIC IPC BX-39A, SIMATIC IPC BX-59A and Virtual Edge Device.

  • Currently only static IP addresses are supported.

  • Each AI Inference Server variant was tested on the following software versions:

    • Industrial Edge platform version ied-os-1.12.0-10-amd64

    • Databus v2.3.2-5

    • External Databus v2.0.1-3

    • IE Flow Creator v1.17.0-2

    • SIMATIC S7 Connector v2.0.1-21

    • Vision Connector v1.0.0

  • The management of the free storage space is not part of AI Inference Server. The user should check regularly that there is enough free storage space to work with a new configuration package file that also contains the AI model.

  • AI Inference Server converts the message data type from the data bus according to the following table. If the data type of an input or output is 'Integer', for example, then the following data bus types can be accepted: 'Int', 'Byte', 'DInt', 'Word', 'LInt', 'SInt', 'USInt', 'UIInt', 'UDInt', 'ULInt', 'DWord', 'LWord', 'Char'.

    Data bus type

    Converted type(how AI Inference Server handles it)

    'Bool'

    'Boolean'

    'Int'

    'Integer'

    'Byte'

    'Integer'

    'DInt'

    'Integer'

    'Real'

    'Double'

    'String'

    'String'

    'Word'

    'Integer'

    'LInt'

    'Integer'

    'SInt'

    'Integer'

    'USInt'

    'Integer'

    'UIInt'

    'Integer'

    'UDInt'

    'Integer'

    'ULInt'

    'Integer'

    'LReal'

    'Double'

    'DWord'

    'Integer'

    'LWord'

    'Integer'

    'Char'

    'Integer'

    'BoolArray'

    'BooleanArray'

    'ByteArray'

    'Int8Array' or 'IntegerArray'

    'SIntArray'

    'Int8Array' or 'IntegerArray'

    'IntArray'

    'Int16Array' or 'IntegerArray'

    'DIntArray'

    'Int32Array' or 'IntegerArray'

    'LIntArray'

    'Int64Array' or' IntegerArray'

    'USIntArray'

    'UInt8Array' or 'IntegerArray'

    'UIntArray'

    'UInt16Array' or 'IntegerArray'

    'UDIntArray'

    'UInt32Array' or 'IntegerArray'

    'ULIntArray'

    'UInt64Array' or 'IntegerArray'

    'WordArray'

    'IntegerArray'

    'DWordArray'

    'UInt32Array' or 'IntegerArray'

    'LWordArray'

    'UInt64Array' or 'IntegerArray'

    'RealArray'

    'Float32Array' or 'DoubleArray'

    'LRealArray'

    'Float64Array' or 'DoubleArray'

    'StringArray'

    'StringArray'

  • AI Inference Server has been tested with machine learning pipelines that were prepared and packaged with AI Model Manager and AI Software Development Kit.

  • AI Inference Server provides Python 3.8.19 and Python 3.10.14 runtime environments.

  • AI Inference Server uses pip 24.0 to install provided Python packages.

  • The state of a Python model is not retained when the user restarts the pipeline: Each run/start of a pipeline via the AI Inference Server user interface is considered a new call of the Python import directive.

  • The performance of the parallel step execution depends significantly on the libraries used, which run on the same CPU core, meaning that the overall performance of the pipeline is only slightly better in some cases.

  • A maximum of one pipeline can be run at the same time with the products with MLFB 6AV2170-0LA10-0AA0 (AI Inference Server) and MLFB 6AV2170-0LA11-0AA0 (AI Inference Server - GPU accelerated - 1 Pipeline)

  • A maximum of three pipelines can be run at the same time with the product with MLFB 6AV2170-0LA10-1AA0 (AI Inference Server - 3 Pipelines)

  • MQTT keep-alive is 60 seconds

  • For performance reasons, AI Inference Server truncates the Python log message created by the user at 1500 characters.