Jetson Orin Nano Hardware Acceleration Issues in Headless Configuration

Hello,

I am requesting technical assistance with hardware video acceleration issues on our NVIDIA Jetson Orin Nano Developer Kit configured for headless operation.

System Configuration:

  • Hardware: NVIDIA Jetson Orin Nano Developer Kit (8GB RAM)
  • JetPack Version: 36.4 (Ubuntu 22.04)
  • Application: Computer vision for industrial license plate detection

Headless Configuration Details: We operate the Jetson remotely without a physical monitor using a dummy display driver configuration:

Section "Device"
    Identifier "DummyDevice"
    Driver "dummy"
    VideoRam 256000
EndSection

Section "Screen"
    Identifier "DummyScreen"
    Device "DummyDevice"
    Monitor "DummyMonitor"
    DefaultDepth 24
    SubSection "Display"
        Depth 24
        Modes "1920x1080_60.0"
    EndSubSection
EndSection

Primary Issues with Headless Operation:

- Hardware Video Acceleration Compatibility:

  • EGL authentication failures: “libEGL warning: DRI2: failed to authenticate”
  • Hardware-accelerated video decoding produces incorrect output (color space issues)

Questions:

  1. What is the recommended headless configuration for Jetson Orin Nano when hardware video acceleration is required?
  2. How can we enable EGL/OpenGL functionality in dummy display environments?
  3. Is there an alternative to dummy driver that maintains remote access while supporting hardware acceleration?
  4. Are there specific Jetson configurations optimized for headless computer vision applications?

Current Impact: We’ve had to implement workarounds that bypass hardware acceleration entirely, significantly reducing performance for our real-time video processing requirements.

Additional Context:

  • System must remain headless for remote operation
  • Requires processing multiple video streams simultaneously
  • GStreamer hardware pipelines fail while software alternatives work
  • Critical for industrial automation application

Thank you for your support.

Hi,
Certain use-cases may not work well in headless setup. Would suggest set up RTSP/UDP to stream out the frame data. There is discussions about Argus and please check
Camera access in headless mode - #9 by DaneLLL
Is it possible to run a GStreamer pipeline in headless mode with the source plugin nvarguscamerasrc?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.