Quick Start
Install
Section titled “Install”# Install uv if you don't have itcurl -LsSf https://astral.sh/uv/install.sh | sh
# Install sparkrun and set up your clusteruvx sparkrun setupThe setup wizard walks you through cluster creation, SSH mesh, and CX7 networking interactively. See the Setup Wizard Walkthrough for a detailed step-by-step guide.
Use a saved cluster
Section titled “Use a saved cluster”Use the cluster created by the wizard as your default. A single-node cluster is
normal; for larger installations, put all nodes in the default and optionally
add smaller logical groups. See Recommended Cluster Setups.
Replace mylab below with the name you chose in the wizard.
sparkrun cluster set-default mylabRun a model
Section titled “Run a model”Let Sparkrun choose placement within the default cluster:
sparkrun run qwen3-1.7b-vllm
# Or select a different saved clustersparkrun run qwen3-1.7b-vllm --cluster mylabsparkrun launches detached containers, follows startup logs, and returns after readiness and required hooks. Use --no-follow to wait quietly, --follow to keep watching after readiness, or --no-ready-wait for launch-only completion. Ctrl+C detaches the observers and leaves the detached workload running. See run.
Run multi-node (tensor parallelism)
Section titled “Run multi-node (tensor parallelism)”With a cluster configured, scale across multiple DGX Sparks:
sparkrun run qwen3-1.7b-vllm --tp 2Each DGX Spark has one GPU with 128 GB unified memory, so --tp 2 splits the model across 2 nodes. See Multi-Node Tensor Parallelism for a full walkthrough.
Inspect a recipe
Section titled “Inspect a recipe”sparkrun show nemotron3-nano-30b-nvfp4-vllmThis displays recipe details and a memory estimate. Unknown or lower-bound estimates do not establish fit; account for runtime allocation, KV cache, and observed host capacity. See Memory-aware fit.
Status, logs, and stop
Section titled “Status, logs, and stop”The following output is illustrative; job IDs, addresses, images, and durations vary.
# Check what's running$ sparkrun status
Job: @sparkrun-transitional/qwen3-1.7b-llama-cpp [e0ed758410a2] (1 container(s)) solo 10.24.11.13 (ib: 192.168.11.13) Up 45 minutes ghcr.io/spark-arena/dgx-llama-cpp:latest logs: sparkrun logs e0ed758410a2 stop: sparkrun stop e0ed758410a2# Re-attach to logs (Ctrl+C is always safe) -- you can reference recipe or job IDsparkrun logs @sparkrun-transitional/qwen3-1.7b-llama-cpp
# Stop a workload (shown here using job ID for unambiguous reference)sparkrun stop e0ed758410a2Spark Arena
Section titled “Spark Arena”Spark Arena is the community hub for DGX Spark recipe benchmarks — browse benchmark results, then run them directly with sparkrun.
Official Recipes
Section titled “Official Recipes”Official Recipes are maintained by the Spark Arena team and hosted on GitHub. They are tested and optimized for NVIDIA DGX Spark systems.
Community Recipes
Section titled “Community Recipes”Community Recipes are contributed by the community and hosted on GitHub.
Next steps
Section titled “Next steps”- Your First Model — step-by-step tutorial
- Recipe Format — write your own recipes
- Registries — add community or private recipe registries
- Networking — CX7 configuration for multi-node clusters