Confirm every item on this checklist before proceeding to . Issues found after installation are harder to diagnose.
Hardware
CPU-only transcoding works for testing and basic validation. Production video nodes need GPU acceleration for reliable throughput.
VRAM determines which AI pipelines the orchestrator can serve. 8 GB runs quantised LLMs only. 24 GB+ unlocks diffusion, audio, vision, and Cascade AI. See for per-model VRAM requirements.
NVENC/NVDEC (video) use dedicated silicon separate from CUDA cores (AI). Both workloads share VRAM. A 24 GB GPU supports video transcoding alongside one warm AI model. See for VRAM coexistence details.
Software
Network
Wallet and Tokens
Plan for hours to days when acquiring LPT, depending on exchange availability and bridging from Ethereum L1.
Quick Verification
Run these commands to confirm the prerequisites are met:
# GPU visible and driver installed
nvidia-smi
# CUDA version (skip if using Docker - it bundles CUDA)
nvcc --version
# Docker GPU passthrough works
docker run --rm --gpus all nvidia/cuda:12.0-base nvidia-smi
# Port 8935 reachable (run from a DIFFERENT machine)
curl -k https://YOUR_PUBLIC_IP:8935/status
Ready
Once all items pass, proceed to installation.
