Orin Nano: Clarification on Maximum Concurrent Virtual Channels

Issue Overview

Users are seeking clarification on the maximum number of concurrent virtual channels supported by the Nvidia Jetson Orin Nano Dev board. The confusion arises from seemingly conflicting information in official documentation:

  • The official summary mentions support for up to 8 cameras via virtual channels.
  • The datasheet (DS-11105-001_v1.1) states that Orin Nano can support up to 16 virtual channels.
  • The same datasheet mentions support for up to six active streams in the Video Input (VI) section.

A specific use case is presented, involving 4x 2-lane MIPI D-PHY configuration with a total of 10 virtual channels across four MIPI CSI ports.

Possible Causes

  1. Documentation Inconsistencies: The discrepancies in official documentation may be due to:

    • Different interpretations of "virtual channels" vs. "cameras"
    • Outdated or erroneous information in some documents
    • Confusion between hardware capabilities and software limitations
  2. Bandwidth Limitations: While the hardware might support more channels, bandwidth constraints could limit the number of concurrent streams.

  3. Software Stack Limitations: The software implementation might impose restrictions on the number of concurrent channels that can be effectively managed.

  4. Processing Power Constraints: The ISP and CPU capabilities might limit the number of concurrent streams that can be processed effectively, especially when considering additional processing like 3A (Auto Exposure, Auto White Balance, Auto Focus).

Troubleshooting Steps, Solutions & Fixes

  1. Understand Hardware Capabilities:

    • The Orin Nano hardware can potentially support up to 16 virtual channels.
    • However, practical limitations may reduce this number in real-world applications.
  2. Consider Bandwidth Limitations:

    • The primary concern from NVIDIA’s perspective is bandwidth usage.
    • Ensure that your total bandwidth requirements are within the Orin Nano’s specifications:
      • MIPI CSI-2: 8 lanes, D-PHY 2.1 (up to 20Gbps)
      • ISP: Maximum frequency of 704 MHz, equivalent to approximately 1408 MPixels
  3. Optimize Virtual Channel Usage:

    • If possible, reduce the number of concurrent virtual channels to 8 or fewer, as this seems to be a more widely supported configuration.
    • For the specific use case mentioned:
      2x MIPI CSI with a single virtual channel on each port
      2x MIPI CSI with up to 4 virtual channels on each port
      

      Consider reducing the number of virtual channels on the latter two ports from 4 to 2, bringing the total to 6 concurrent channels.

  4. Evaluate Software Stack Requirements:

    • Be aware that software limitations may restrict the number of concurrent cameras to 8, even if the hardware supports more.
    • Consider the processing requirements of additional software components like 3A (Auto Exposure, Auto White Balance, Auto Focus).
  5. Perform Practical Tests:

    • Once you have the hardware, conduct thorough tests to determine the actual limitations in your specific use case.
    • Monitor system performance, including CPU usage, memory consumption, and thermal characteristics, while running multiple concurrent streams.
  6. Consider Alternative Hardware:

    • If your requirements exceed the capabilities of the Orin Nano, consider using more powerful alternatives like the AGX Orin or AGX Xavier.
  7. Consult NVIDIA Support:

    • For definitive answers and the most up-to-date information, reach out to NVIDIA’s official support channels.
    • Provide detailed information about your specific use case, including:
      • Number of physical cameras
      • Number of virtual channels
      • Resolution and frame rate of each stream
      • Any additional processing requirements (e.g., ISP, 3A)
  8. Stay Updated:

    • Keep an eye on NVIDIA’s official documentation and forums for any updates or clarifications regarding virtual channel support on the Orin Nano.

By following these steps and considering the various factors involved, you should be able to determine the most appropriate configuration for your Orin Nano implementation and maximize the number of concurrent virtual channels within the system’s capabilities.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *