Skip to content

ARM-based ECU Configuration#

Deploy Autoware on ARM platforms, focusing on NVIDIA Jetson and AGX Orin for LSA vehicles.

System Preparation#

CUDA Toolkit Installation#

For NVIDIA AGX Orin#

Important: Do NOT install CUDA packages manually on AGX Orin as it may break the system. CUDA is included with JetPack and must be installed through NVIDIA's SDK Manager.

Installation Steps:

  1. Download NVIDIA SDK Manager from https://developer.nvidia.com/nvidia-sdk-manager

  2. Select JetPack 6.0 (or latest version) in SDK Manager

  3. This will flash the Orin device and install the appropriate CUDA version

  4. JetPack 6.0 includes CUDA 12.2, cuDNN, TensorRT, and other essential libraries

  5. Flash and Install

  6. Connect your AGX Orin to the host PC via USB-C
  7. Follow SDK Manager prompts to flash the device

  8. The process will install Ubuntu, CUDA, and all necessary drivers

  9. Verify Installation after flashing: 

    # Check CUDA version
nvcc --version

# Verify Jetson platform and monitor system
sudo apt install -y python3-pip
pip3 install jetson-stats
sudo jtop

# Check JetPack version
cat /etc/nv_tegra_release

For Other ARM64 Platforms#

For ARM64 platforms other than NVIDIA Jetson: - Check the manufacturer's product specifications or manual for CUDA support - Most non-NVIDIA ARM platforms do not support CUDA - Consider using CPU-only or other acceleration options if CUDA is not available

Supported JetPack Versions by Platform: - AGX Orin: JetPack 6.0 or later (recommended) - Xavier Series: JetPack 5.1 or later - Nano/TX2: Check NVIDIA's compatibility matrix

Platform-Specific Optimizations#

NVIDIA AGX Orin Configuration#

Power Management#

Configure power modes based on deployment requirements:

# Development mode - Maximum performance
sudo nvpmodel -m 0  # MAXN mode
sudo jetson_clocks

# Production mode - Balanced performance/efficiency
sudo nvpmodel -m 1  # 30W mode
sudo jetson_clocks --restore

Memory Configuration#

Optimize memory allocation for Autoware workloads:

# Increase GPU memory allocation
echo "gpu_mem_size=8G" | sudo tee /etc/modprobe.d/tegra.conf

# Configure swap for memory-intensive operations
sudo fallocate -l 16G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile

Hardware Acceleration Setup#

Enable DLA (Deep Learning Accelerator)#

# ansible/roles/agx_orin_dla/tasks/main.yml
---
- name: Enable DLA cores
  lineinfile:
    path: /etc/environment
    line: "{{ item }}"
  loop:
    - 'CUDA_VISIBLE_DEVICES=0'
    - 'DLA_VISIBLE_DEVICES=0,1'
    - 'TF_ENABLE_TENSORRT_DLA=1'

Performance Tuning#

CPU Governor Settings#

# Set performance governor for all cores
for cpu in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor; do
  echo performance | sudo tee $cpu
done

# Make persistent
echo 'GOVERNOR="performance"' | sudo tee /etc/default/cpufrequtils

GPU Optimization#

# ansible/roles/gpu_optimization/tasks/main.yml
---
- name: Set GPU clock to maximum
  command: nvidia-smi -pm 1

- name: Configure GPU memory growth
  lineinfile:
    path: /etc/environment
    regexp: '^TF_FORCE_GPU_ALLOW_GROWTH='
    line: 'TF_FORCE_GPU_ALLOW_GROWTH=true'

- name: Set CUDA device order
  lineinfile:
    path: /etc/environment
    regexp: '^CUDA_DEVICE_ORDER='
    line: 'CUDA_DEVICE_ORDER=PCI_BUS_ID'

Storage Optimization#

NVMe Configuration for High-Speed Logging#

# ansible/roles/storage_optimization/tasks/main.yml
---
- name: Configure NVMe for optimal performance
  lineinfile:
    path: /etc/fstab
    line: '/dev/nvme0n1p1 /var/log/autoware ext4 noatime,nodiratime,nobarrier 0 2'

- name: Set up log rotation for Autoware
  template:
    src: autoware_logrotate.j2
    dest: /etc/logrotate.d/autoware

SD Card Optimization (Jetson Nano/Xavier NX)#

# Reduce SD card wear
echo "vm.swappiness=10" | sudo tee -a /etc/sysctl.conf
echo "vm.vfs_cache_pressure=50" | sudo tee -a /etc/sysctl.conf

# Move temporary files to RAM
echo "tmpfs /tmp tmpfs defaults,noatime,mode=1777 0 0" | sudo tee -a /etc/fstab

Monitoring and Debugging#

Jetson-Specific Monitoring#

# Real-time system monitoring
sudo jtop

# GPU/CPU/Memory stats
tegrastats

# Temperature monitoring
cat /sys/devices/virtual/thermal/thermal_zone*/temp

Performance Profiling#

# Profile Autoware with Nsight Systems
nsys profile -t cuda,nvtx,osrt,cudnn,cublas \
  -o autoware_profile \
  ros2 launch autoware_launch autoware.launch.xml

# Analyze with Nsight Compute
ncu --target-processes all \
  --metrics gpu__time_duration.sum \
  ros2 run perception_node perception_node

Next Steps#

See Sensor Configuration Guide for detailed sensor setup