nvidia-datacenter-bringup

# nvidia-datacenter-bringup Opinionated greenfield recipe for **NVIDIA datacenter GPUs on Ubuntu 24.04 LTS** — get from a clean OS install to a healthy host where `nvidia-smi` reports all GPUs, `nvidia-fabricmanager` is `active (running)`, and the gpu-operator `cuda-validator` pod passes. Air-gap is the primary case; connected sites use the same packages from the same upstream URLs. **Scope boundary.** This skill ends at a *healthy* host, not a *fast* one. Once `cuda-validator` passes, performance work — IOMMU/hugepages/NUMA pinning, PCIe ACS, persistence mode, the pinned-host↔GPU memcpy ceiling — belongs to the `gpu-host-tuning` skill (same `inference-host` plugin). Do not tune here. ## Decision tree | Question | Answer | Read | |---|---|---| | Has Blackwell silicon (B300/B200/B100)? | Yes | Open kernel modules **mandatory** — proprietary is unsupported [[open-modules-transition]] | | Grace Hopper (GH200)? | Yes | Open kernel modules **mandatory** (same as Blackwell). Otherwise 8-GPU SXM path [[hopper-recipe]] | | HGX 8-GPU SXM with 3rd-gen NVSwitch (H100/H200/H800 in XE9680 or similar)? | Yes | Open recommended (not mandatory). Use `cuda-drivers-fabricmanager-<branch>` meta; **skip** `nvlink5-<branch>`, NVLSM, DOCA-OFED entirely. Min driver 525+ for H100, 535+ for H200 [[hopper-recipe]] | | HGX 4-GPU SXM (H100 in XE8640, A100 4-GPU)? | Yes | **No NVSwitch on this baseboard** — direct NVLink mesh between 4 GPUs. **Skip fabricmanager entirely** + DOCA + NVLSM. Three-packa...