System Requirements

Redact-Edge is built on GStreamer and NVIDIA DeepStream , which means it relies heavily on NVIDIA GPUs for efficient processing. To ensure optimal performance and compatibility, please make sure your system meets the following requirements before running Redact-Edge.

System Requirements

Redact RT is built on GStreamer and NVIDIA DeepStream. It relies on NVIDIA GPUs for efficient processing.

Before running Redact RT, ensure that your system meets the requirements for the relevant architecture.


x86_64

Hardware Requirements

ComponentRequirement
GPUNVIDIA GPU
GPU memoryMinimum 4 GiB
Compute capability7.5, 8.0, 8.6, 8.9, 9.0, or 12.0
System memoryMinimum 4 GB RAM
Disk spaceMinimum 15 GB

To check the GPU model and compute capability, run:

nvidia-smi --query-gpu=name,compute_cap --format=csv

Software Requirements

  • Ubuntu 20.04 or later
  • NVIDIA driver version R570.133.20 or later, required for DeepStream 8.0
  • Docker with NVIDIA configured as the default runtime
  • NVIDIA Container Toolkit

Verify GPU Visibility via CDI

The NVIDIA Container Toolkit uses the Container Device Interface (CDI) to expose GPUs to containers.

Verify that all installed GPUs are visible through CDI:

nvidia-ctk cdi list | grep -E '^nvidia\.com/gpu=[0-9]+$'

The listed GPU indices must match the GPU indices available on the host.

If one or more GPUs are missing, regenerate the CDI specification:

sudo nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml

aarch64

Supported NVIDIA Jetson Devices

Redact RT officially supports the following devices:

  • NVIDIA Jetson AGX Orin
  • NVIDIA Jetson Orin NX

Other Jetson models, such as the Jetson Orin Nano, may work but are not officially supported.

Hardware Requirements

ComponentRequirement
Compute capability8.7
System and GPU memoryMinimum 4 GB
Disk spaceMinimum 11 GB

Software Requirements

  • JetPack 6.1 or later
    • JetPack 6.2.2 is recommended
    • JetPack includes the NVIDIA R560-series driver required by DeepStream 7.1 on Jetson
  • Ubuntu 18.04 or later
  • Docker with NVIDIA configured as the default runtime
  • NVIDIA Container Toolkit

Did this page help you?