Installation and Prerequisites

This guide provides instructions for installing system dependencies, updating NPU drivers, compiling the NERVI client, and initializing the Python backend.

💻 System Requirements

Supported Hardware

Processor: Intel Core Ultra (Generation 1 or newer) containing an integrated Intel NPU.
Accelerator: Intel AI Boost (integrated Neural Processing Unit).
RAM: Minimum 16GB (32GB recommended for running larger models like 7B parameters).

Software Requirements

Operating System: Windows 11 (64-bit) or modern Linux distribution.
Go compiler: Go 1.21 or higher.
Python: Python 3.8 to 3.11 (must be added to your system PATH).
NPU Driver: Intel NPU Driver version 31.0.100.1234 or higher.

⚙️ Driver Setup (Intel NPU)

Before running local models, you must verify that your processor's NPU driver is installed and active.

Verifying NPU Driver on Windows

Open the Device Manager (right-click the Start Menu and select Device Manager).
Expand the Intel(R) AI Boost category.
Right-click Intel(R) AI Boost and choose Properties.
Check the Driver Version tab. Ensure it is updated.

🛠️ Building the Go CLI Client

Clone the repository:

git clone https://github.com/dsainvg/NERVI.git
cd NERVI

Compile the binary:

Windows:
```
go build -o nervi.exe main.go
```
Linux/Unix:
```
go build -o nervi main.go
```

🤖 Automatic Environment Scaffolding

NERVI uses a self-contained environment strategy to prevent polluting your system-wide Python installation.

On the first invocation of any model-related CLI command, the Go client will execute the setup sequence:

Folder Scaffold: Creates configuration folders under %AppData%\Roaming\nervi.
Virtual Environment: Runs python -m venv venv inside the configuration directory.
Pip Installation: Installs npuserver from PyPI, pulling in openvino-genai, flask, and huggingface-hub.
Sentinel Lock: Writes a .nervi_ready sentinel. Subsequent commands skip the install checks.

🔄 Upgrading the Python Backend

To update the Python backend package:

./nervi update