Skip to main content

Prerequisites

This document describes the prerequisites for deploying the BTT InferGrid miner-side components. The miner-side stack is designed for one Linux host with NVIDIA GPUs.

Host Requirements

RequirementDescription
Linux x86_64Host operating system
NVIDIA driverInstalled and working
nvidia-smiGPU driver management tool
DockerAvailable to the current operator
Docker ComposeCompose v2 plugin available as docker compose
NVIDIA Container ToolkitDocker GPU support
Python3.10+
uvRecommended development and execution tool
note

This documentation does not cover full NVIDIA driver installation. Driver installation and basic GPU visibility remain the host administrator's responsibility.

Install Docker Compose

miner-cli renders Docker Compose deployments, but Docker Compose must be installed manually. Install the Compose v2 plugin with your OS package manager or Docker's official instructions, then verify:

docker compose version

Docker, Docker Compose, NVIDIA drivers, and NVIDIA Container Toolkit installation usually require root or sudo. Run miner-cli and keep YAML files, model cache, and persistent miner directories such as /data/minerhome under a normal operator account, not under /root.

Install uv

uv is a fast Python package manager:

# Method 1: Official install script
curl -LsSf https://astral.sh/uv/install.sh | sh

# Method 2: pip install
pip install uv

Verify installation:

uv --version

Use miner-cli

Method 1: Run from source

cd miner-cli
uv sync
uv run miner-cli doctor

Method 2: Install to current Python environment

cd miner-cli
pip install .
miner-cli doctor

Development Checks

Both projects use Python package entrypoints and pytest-based tests:

# Install dev dependencies and run tests
uv sync --extra dev
uv run pytest

# Code linting
uv run --extra dev ruff check .
tip

Run these commands from each project's directory.

Next Steps

Once the environment is verified, proceed to Quick Start.