Neuronova SDK — Early Access

This is an early-access release of the Neuronova SDK, designed to give you hands-on access to a new computing paradigm: always-on, ultra-low-power analog AI at the edge.

Some features may be incomplete, APIs may evolve, and parts of the experience will change as the technology matures. This is expected at this stage.

We’re releasing the SDK early on purpose. We believe the fastest way to build the right platform is to build it together with the people who are pushing real application to the real world, not in isolation.

Your feedback is extremely important to us. What works, what doesn’t, what feels missing, all of it directly influences our roadmap, our tooling, and the next iterations of both software and silicon.

This is not just about testing an SDK.
It’s about shaping a new always-on intelligence layer at the edge.

Thank for being with us!

OK

Development Environment Setup (Required)

To build, develop, and ship NWAVE consistently across multiple GPU platforms, we provide preconfigured Docker environments for developing PyTorch extensions with HIP, supporting both:

• AMD GPUs (ROCm)
• NVIDIA GPUs (CUDA)
• No GPU setups (cpu-only) (with limited features)

These environments ensure:

• Identical toolchains across platforms
• Consistent ROCm / CUDA / HIP versions
• Proper GPU access and permissions
• A reproducible setup for all developers

Warning

Using these environments is strongly recommended. Deviating from them will likely result in build issues, missing symbols, or platform-specific incompatibilities.

The Dockerfiles are derived from the official ROCm examples repository, and are available in the downloaded zip file for completeness.

---

What’s Included

Each environment comes fully set up with:

• Ubuntu base image
• HIP development tools and libraries
• Support for ROCm (AMD)/CUDA (NVIDIA)/CPU-only
• Anaconda Python distribution
• A ready-to-use Conda environment (PyTorch)
• CMake and standard development tools
• Correct GPU device access, IPC, and shared memory configuration

---

Prerequisites

Before building or running the containers, make sure you have:

• Docker installed
• Linux based OS (or WSL if using windows)

• If you have GPU:

  • AMD users:

    • ROCm-compatible GPU
    • ROCm drivers installed on the host

  • NVIDIA users:

    • CUDA-compatible GPU
    • NVIDIA drivers installed on the host

• The extracted NWAVE package containing:

  • assets/downloads/nwave-100a/install.sh
  • wheel files and Dockerfile templates in the same folder

Mac-OS is not tested due to limited hardware supports, feel free to test the CPU-only environments on mac machines!

---

Automated Installation

Before starting the installation, you should create a folder named DockerMem in a filesystem position that is easily accessible (for instance in /home/<user>). This is the folder in which all files/projects must be be placed in order to be used in the container (and therefore by NWAVE too).

All environment setup steps are automated through install.sh. From the package folder:

cd nwave-100a
chmod +x install.sh

After that, in both the GPU vendor case, an image name and a container name must be chosen to launch the installer (any name is ok, as long as it doesn't contains - or /).

AMD (ROCm)

./install.sh \
  --vendor amd \
  --dockermem-dir /absolute/path/to/DockerMem \
  --image-name kernel_devel_amd_630 \
  --container-name nwave_amd_dev \
  --amd-gfx 11.0.0

Use --amd-gfx 10.3.0 for RDNA/RDNA2 when needed.

NVIDIA (CUDA)

./install.sh \
  --vendor nvidia \
  --dockermem-dir /absolute/path/to/DockerMem \
  --image-name kernel_devel_nvidia_630 \
  --container-name nwave_nvidia_dev

CPU-only

./install.sh \
  --vendor cpu \
  --dockermem-dir /absolute/path/to/DockerMem \
  --image-name kernel_devel_cpu_630 \
  --container-name nwave_cpu_dev

---

Using the Environment

NWAVE runs inside a Docker container.
A container is like a ready-to-use environment with everything already installed (Python, PyTorch, etc.).

---

Using the Environment (Terminal)

Start or enter the container

If it’s already running:

docker attach {your_container_name}

If it’s not running:

docker start -ai {your_container_name}

If you used the install script, the name is usually: nwave_{vendor}_dev

---

Activate the environment

Once inside, run:

source activate PyTorch

⚠️ You must run this every time you enter the container.

---

Using the Environment in VSCode

Click the bottom-left corner in VSCode:

Then select:

"Attach to Running Container..."

Choose your container from the list.

---

If no container appears

Start it first from the terminal:

docker start -ai {your_container_name}

---

Notes

• Select the PyTorch environment in VSCode
• Files are available under /home/workspaces that mirrors the host folder /home/{your-pc-name}/DockerMem
• Install the Jupyter extension if you want to run notebooks

---

Troubleshooting

GPU Not Detected

• AMD: Verify ROCm installation and presence of /dev/kfd and /dev/dri
• NVIDIA: Verify NVIDIA Container Toolkit and driver installation

Permission Issues

• Ensure your user is in the docker group
• Confirm UID/GID match between host and container
• Check permissions on the shared workspace directory

---

Contributing

For issues, feature requests, or questions, either raise a github issue on the release repo or send a mail at giuseppe@neuronovatech.com.