Skip to content

Repository files navigation

NVIDIA DGX: At-Home AI Stack

Self-hosted inference and apps on one or two Nvidia DGX Spark nodes (arm64 / GB10 / Ubuntu 24.04).

Out of the box

  1. Unbox the Spark. Cable mgmt LAN and, if you have two, the 200 Gb/s DAC.
  2. Install a coding agent first (Grok Build, Codex, Claude Code, or similar).
  3. Feed it llms.txt and llms-full.txt (human page: llms.html).
  4. Answer the interview. The agent builds clustered TP=2 or isolated solo TP=1 from current eugr/spark-vllm-docker commands (vllm-node, no --tf5, hf-download.sh, no-Ray default).

The Pages root also has a tabbed generator that downloads the same procedure as setup-node-a.sh / setup-node-b.sh.

Architecture A — clustered Architecture B — isolated solo
vLLM One process, TP=2 over the DAC One process per node, TP=1
GPU Shared pool Each node owns its GB10
Who sees prompts Head API sees both UIs Each API sees only its UI
Typical models 122B-class FP8 (often cluster-only) INT4 122B-class and/or 35B FP8
When Two nodes, one shared model One Spark, isolation, or independent models

Documentation names: node-a (personal + Hermes) and node-b (workload UI). Documentation addresses only: 192.0.2.21 / 192.0.2.22 and 198.51.100.0/30 — never configure those on a real box.

Port 8000 must not be published on any overlay. There is no LiteLLM proxy. Templates live in templates/.

Contributors

Languages