Linux Setup

Configure your Linux environment for high-performance inference tracking. This guide covers standard x86_64 architectures and ARM64 platforms, including NVIDIA Jetson devices running JetPack.

download

Core Installation

Install the primary AIBench package via pip. Ensure you are operating within a virtual environment to avoid dependency conflicts.

$ pip install aibench-cli

verified

Verification

Run the following to confirm installation and check available hardware backends.

$ python -m aibench scan

memory

Supported Hardware Backends

BACKEND	ARCHITECTURE	STATUS	REQUIRES PACKAGE
ONNX CPU	x86_64 / ARM64	check_circle	`onnxruntime`
CUDA	x86_64	check_circle	`onnxruntime-gpu`
PyTorch (CPU)	x86_64 / ARM64	check_circle	`torch`
PyTorch (CUDA)	x86_64 / ARM64	check_circle	`torch (CUDA build)`
OpenVINO	x86_64	check_circle	`openvino`
TFLite	x86_64 / ARM64	check_circle	`tflite-runtime`

developer_board

NVIDIA Jetson (ARM64) Configuration

JetPack 5.x / 6.x

Deploying AIBench on edge devices like the Jetson Orin series requires specific system-level dependencies before installing the Python package. Ensure your JetPack installation includes standard CUDA toolkits.

1. System Dependencies

$ sudo apt-get update
$ sudo apt-get install -y \
    libhdf5-serial-dev \
    libopenblas-dev

2. Jetson PyTorch Wheel

# Install NVIDIA's Jetson PyTorch:
$ pip3 install torch --index-url \
    https://pypi.jetson-ai-lab.io/jp6/cu126