Provider Setup Guide

Turn your idle GPUs into a revenue stream. VexNode providers earn crypto-native rewards on Bittensor for every GPU-hour served. This guide walks you through setup from bare metal to live.

1. System Requirements

NVIDIA GPU with 16 GB+ VRAM (RTX 3090, RTX 4090, A100, H100)
Ubuntu 22.04 LTS or later (other Linux distros supported)
NVIDIA Driver 535+ and CUDA 12.1+
Docker Engine 24+ with NVIDIA Container Toolkit
Stable internet connection (100 Mbps+ recommended)

2. Install the VexNode Agent

The agent runs as a background service, manages job execution in isolated containers, and reports telemetry to the VexNode orchestration layer.

bash

curl -fsSL https://get.vexnode.ai/agent | bash
sudo systemctl enable --now vexnode-agent

3. Configuration

After installation, configure your provider identity and payout preferences:

/etc/vexnode/agent.yaml

provider:
  name: "my-gpu-rig"
  region: "us-east"

wallet:
  type: "bittensor"
  address: "5F3sa2TJbN1..."

gpus:
  - device: 0
    max_concurrent_jobs: 2
  - device: 1
    max_concurrent_jobs: 2

network:
  port: 8472
  advertise_bandwidth: 1000  # Mbps

4. Benchmarking

Run the built-in benchmark to verify your hardware and establish a performance baseline. Results are submitted to the network and determine your tier placement and job routing priority.

bash

vexnode benchmark --all

# Example output:
# GPU 0 (RTX 4090): 847 TFLOPS FP16 — Tier: Pro
# GPU 1 (RTX 4090): 843 TFLOPS FP16 — Tier: Pro
# Network latency: 12ms avg to orchestrator
# Status: PASSED — eligible for job routing

5. Going Live

Once benchmarking passes, your node is automatically registered on the network and begins receiving job assignments. Monitor your node status and earnings from the provider dashboard:

bash

# Check agent status
vexnode status

# View real-time logs
journalctl -u vexnode-agent -f

# Check earnings
vexnode earnings --period month

Payouts are processed daily to your configured wallet. You can track earnings, uptime, and job history in the Provider Dashboard.

3. Configuration

After installation, configure your provider identity and payout preferences:

/etc/vexnode/agent.yaml

provider:
  name: "my-gpu-rig"
  region: "us-east"

wallet:
  type: "bittensor"
  address: "5F3sa2TJbN1..."

gpus:
  - device: 0
    max_concurrent_jobs: 2
  - device: 1
    max_concurrent_jobs: 2

network:
  port: 8472
  advertise_bandwidth: 1000  # Mbps

4. Benchmarking

Run the built-in benchmark to verify your hardware and establish a performance baseline. Results are submitted to the network and determine your tier placement and job routing priority.

bash

vexnode benchmark --all

# Example output:
# GPU 0 (RTX 4090): 847 TFLOPS FP16 — Tier: Pro
# GPU 1 (RTX 4090): 843 TFLOPS FP16 — Tier: Pro
# Network latency: 12ms avg to orchestrator
# Status: PASSED — eligible for job routing

5. Going Live

Once benchmarking passes, your node is automatically registered on the network and begins receiving job assignments. Monitor your node status and earnings from the provider dashboard:

bash

# Check agent status
vexnode status

# View real-time logs
journalctl -u vexnode-agent -f

# Check earnings
vexnode earnings --period month

Payouts are processed daily to your configured wallet. You can track earnings, uptime, and job history in the Provider Dashboard.