[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-setup-nvidia-gpu-host":3,"mdc--c6o82h-key":34,"related-org-nvidia-tao-setup-nvidia-gpu-host":2099,"related-repo-nvidia-tao-setup-nvidia-gpu-host":2258},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"tao-setup-nvidia-gpu-host","configure NVIDIA GPU host environments","Host setup for TAO GPU backends. Checks and, after user approval, installs NVIDIA driver branch 580, CUDA Toolkit 13.0, and NVIDIA Container Toolkit 1.19.0 for Docker\u002Flocal-Docker and Kubernetes GPU worker hosts. The `--check-only` path works on any Linux distribution; `--install` automates debian-family (Ubuntu\u002FDebian\u002FPop!_OS\u002FMint\u002FZorin\u002FRaspbian), rhel-family (Fedora\u002FRHEL\u002FRocky\u002FAlmaLinux), and suse-family (openSUSE\u002FSLES) hosts, and prints actionable manual-install steps for everything else. Use when the user asks to \"set up an NVIDIA GPU host\", \"check TAO Docker GPU runtime\", or prepare a Kubernetes GPU worker for TAO.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,14,17,20],{"name":9,"slug":8,"type":13},"tag",{"name":15,"slug":16,"type":13},"Docker","docker",{"name":18,"slug":19,"type":13},"Kubernetes","kubernetes",{"name":21,"slug":22,"type":13},"Infrastructure","infrastructure",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:28:20.893503","Apache-2.0",281,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftao-setup-nvidia-gpu-host","---\nname: tao-setup-nvidia-gpu-host\ndescription: >-\n  Host setup for TAO GPU backends. Checks and, after user approval, installs\n  NVIDIA driver branch 580, CUDA Toolkit 13.0, and NVIDIA Container Toolkit\n  1.19.0 for Docker\u002Flocal-Docker and Kubernetes GPU worker hosts. The\n  `--check-only` path works on any Linux distribution; `--install` automates\n  debian-family (Ubuntu\u002FDebian\u002FPop!_OS\u002FMint\u002FZorin\u002FRaspbian), rhel-family\n  (Fedora\u002FRHEL\u002FRocky\u002FAlmaLinux), and suse-family (openSUSE\u002FSLES) hosts, and\n  prints actionable manual-install steps for everything else. Use when the user\n  asks to \"set up an NVIDIA GPU host\", \"check TAO Docker GPU runtime\", or\n  prepare a Kubernetes GPU worker for TAO.\nlicense: Apache-2.0\ncompatibility: Runs `--check-only` on any Linux distribution. `--install` automates Ubuntu 22.04\u002F24.04 + Debian 12 (apt), Fedora + RHEL\u002FRocky\u002FAlmaLinux 9\u002F10 (dnf), and openSUSE Leap \u002F SLES 15 (zypper). Requires sudo\u002Froot, internet access to NVIDIA package repositories (and download.docker.com on rhel-family), and an x86_64 or aarch64 (sbsa) host. Other distributions (Arch, Alpine, Gentoo, NixOS, …) get a clear error that names the version targets and the NVIDIA install-guide URL.\nmetadata:\n  author: NVIDIA Corporation\n  version: \"0.1.0\"\nallowed-tools: Read Bash\ntags:\n- setup\n- nvidia\n- cuda\n- docker\n- kubernetes\n---\n\n# NVIDIA GPU Host Setup\n\nUse this setup skill before TAO workflows run on the `docker`, `local-docker`,\nor `kubernetes` backend. It standardizes the host GPU runtime on:\n\n- NVIDIA driver branch `580` (open kernel module preferred)\n- CUDA Toolkit package `cuda-toolkit-13-0`\n- NVIDIA Container Toolkit `1.19.0`\n- Docker engine — only installed for `docker` \u002F `local-docker` backends and\n  only when Docker is missing. The package picked depends on the distro\n  family (`docker.io` on Debian-family by default, `moby-engine` \u002F\n  `docker-ce` from `download.docker.com` on RHEL-family, `docker` on\n  SUSE-family). Pass `--skip-docker-install` to opt out.\n\nThe check is safe and read-only by default — it works on any Linux\ndistribution because it only probes `nvidia-smi`, the CUDA toolkit path,\nthe installed container-toolkit package version (via `dpkg`\u002F`rpm`\u002Fthe\n`nvidia-ctk` binary version), and the Docker daemon's NVIDIA runtime.\n\nInstallation must be explicitly authorized by the user and rerun with\n`--install`. The install path is automated for these distro families:\n\n| Family | Tested distros | Manager | Notes |\n|---|---|---|---|\n| debian | Ubuntu 22.04 \u002F 24.04, Debian 12 (and derivatives Pop!_OS, Mint, Zorin, Raspbian, KDE Neon, etc. via `UBUNTU_CODENAME` \u002F `VERSION_CODENAME`) | `apt-get` | Adds NVIDIA `cuda-keyring` + Container Toolkit `.list`. Docker via `docker.io` (override `$DOCKER_PACKAGE_DEBIAN`). |\n| rhel | Fedora 39+, RHEL \u002F Rocky \u002F AlmaLinux 9 and 10 | `dnf` (or `yum`) | Adds NVIDIA `cuda-\u003Cdistro>.repo` + Container Toolkit `.repo`. Docker via Fedora `moby-engine` when available, otherwise `docker-ce` from `download.docker.com`. |\n| suse | openSUSE Leap 15, SLES 15 | `zypper` | Adds the same NVIDIA `.repo` files. Docker via the distribution `docker` package. |\n| other (Arch, Alpine, Gentoo, NixOS, FreeBSD, …) | n\u002Fa | n\u002Fa | `--install` exits with a clear error listing the version targets and the NVIDIA install-guide URLs. Install manually, then rerun `--check-only`. |\n\n## Quick Start\n\nFrom the skill bank root:\n\n```bash\n# Check the local Docker backend host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n\n# Install or repair after user approval.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --install\n\n# Check a Kubernetes GPU worker host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend kubernetes --check-only\n```\n\n> ⚠️ **Note — running non-interactively (agent\u002Fskill runs):** a skill run has no terminal, so the\n> installer's `Continue? [y\u002FN]` prompt cannot be answered. After running `--check-only` to preview and\n> getting the user's approval, append the assume-yes flag (`--yes`) to the `--install` command so it\n> proceeds without a prompt — this auto-confirms installation of system packages (NVIDIA driver, CUDA\n> Toolkit, NVIDIA Container Toolkit, and Docker for Docker backends) and modifies the host, so only do\n> this on a host you control. A person running `--install` directly at a terminal gets the prompt instead.\n\n## Workflow Contract\n\nDocker and Kubernetes workflows must run the check before submitting GPU work:\n\n```bash\nSETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n\nbash \"$SETUP_SCRIPT\" --backend docker --check-only || {\n  echo \"MISSING: TAO GPU host runtime is not ready.\"\n  echo \"After user approval, run (append --yes for non-interactive agent runs):\"\n  echo \"  bash \\\"$SETUP_SCRIPT\\\" --backend docker --install\"\n  exit 1\n}\n```\n\nNever install silently. If the check fails, explain what is missing, ask the\nuser to authorize the fix, then run the install command and rerun the check.\n\n## What The Installer Does\n\nThe installer dispatches on the detected distribution family. On every\nsupported family it adds NVIDIA's CUDA and Container Toolkit repositories\n(if missing), installs the pinned runtime packages, optionally installs\nDocker, wires the NVIDIA Docker runtime, and adds the invoking user to\nthe `docker` group.\n\nCommon steps (all families):\n\n1. Adds NVIDIA's CUDA repository if missing (apt `cuda-keyring` deb,\n   `cuda-\u003Cdistro>.repo` for dnf\u002Fzypper).\n2. Adds NVIDIA's Container Toolkit repository if missing (`.list` for apt,\n   `.repo` for dnf\u002Fzypper).\n3. Installs the matching kernel header \u002F devel package for the running\n   kernel.\n4. Installs the driver branch 580 packages, `cuda-toolkit-13-0`, and the\n   Container Toolkit pinned to `1.19.0` (the dpkg-suffixed `1.19.0-1` is\n   the same upstream version expressed for apt).\n5. For Docker backends and when Docker is missing, installs Docker\n   (override \u002F opt-out flags below), enables\u002Fstarts the daemon, then runs\n   `nvidia-ctk runtime configure --runtime=docker` and restarts Docker\n   when `systemctl` is available.\n6. Adds the invoking user (`$SUDO_USER` if available, else `$USER`) to the\n   `docker` group so subsequent shells can run `docker` without `sudo` —\n   opt out with `--skip-docker-group`. **The new group membership does not\n   take effect in the current shell**: log out and back in, or run\n   `newgrp docker` in each new shell.\n7. Attempts `modprobe nvidia` so verification can pass before reboot.\n\nFamily-specific package selections:\n\n| Step | debian-family | rhel-family | suse-family |\n|---|---|---|---|\n| Kernel headers | `linux-headers-$(uname -r)` | `kernel-devel-$(uname -r)`, `kernel-headers-$(uname -r)` | `kernel-default-devel` |\n| Driver | `nvidia-driver-pinning-580`, `nvidia-open-580` (override: `$NVIDIA_DRIVER_PACKAGE_DEBIAN`) | `nvidia-driver-cuda`, `kmod-nvidia-open-dkms` (override: `$NVIDIA_DRIVER_PACKAGE_RHEL`, `$NVIDIA_DRIVER_KMOD_RHEL`) | `nvidia-open-driver-G06-signed-kmp-default` (override: `$NVIDIA_DRIVER_PACKAGE_SUSE`) |\n| CUDA toolkit | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` |\n| Container Toolkit | `nvidia-container-toolkit=1.19.0-1` + base\u002Ftools\u002Flibs | `nvidia-container-toolkit-1.19.0` + base\u002Ftools\u002Flibs | same as rhel |\n| Docker | `docker.io` (override: `$DOCKER_PACKAGE_DEBIAN`) | `moby-engine`+`moby-cli` on Fedora when available, else `docker-ce docker-ce-cli containerd.io` from `download.docker.com` | `docker` |\n\n## Verification\n\nAfter installation, verify:\n\n```bash\nnvidia-smi\n\u002Fusr\u002Flocal\u002Fcuda-13.0\u002Fbin\u002Fnvcc --version\ndocker info --format '{{json .Runtimes}}' | grep nvidia\nsudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smi\n```\n\nExpected `nvidia-smi` output includes driver `580.x` and CUDA Version `13.0`.\nExpected `nvcc` output includes `release 13.0`.\n\n## Kubernetes Notes\n\nFor self-managed Kubernetes clusters, run the host installer on every GPU\nworker node or bake the same package set into the node image before installing\nthe NVIDIA GPU Operator or device plugin.\n\nThe workflow check also warns if `kubectl` is available but the cluster reports\nno `nvidia.com\u002Fgpu` allocatable capacity. In that case, install\u002Fconfigure the\nNVIDIA GPU Operator after the worker host runtime is ready:\n\n```bash\nhelm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\nhelm install --wait gpu-operator -n gpu-operator --create-namespace nvidia\u002Fgpu-operator\n```\n\nManaged Kubernetes providers may own driver installation through node images or\nGPU Operator policy. Do not overwrite a provider-managed GPU node without user\napproval and a rollback plan.\n\n## Failure Modes\n\n**Unsupported distribution family**: `--install` automates debian-, rhel-,\nand suse-family hosts. On Arch, Alpine, Gentoo, NixOS, FreeBSD, or anything\nwithout `\u002Fetc\u002Fos-release` (e.g. macOS), the script exits with a clear error\nthat lists the four version targets and the upstream NVIDIA install-guide\nURLs:\n\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fcuda-installation-guide-linux\u002F`\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fdatacenter\u002Fcloud-native\u002Fcontainer-toolkit\u002Flatest\u002Finstall-guide.html`\n- `https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F`\n\nInstall those four pieces using your distribution's package manager and\nrerun the script with `--check-only` to verify. The check is universally\nportable — it only queries the binaries \u002F package databases — so once the\nruntime is in place the workflow contract is satisfied regardless of the\nunderlying distro.\n\n**Unsupported Ubuntu\u002FDebian derivative**: When `ID` is e.g. `pop`, `mint`,\n`zorin`, `raspbian`, or another debian-family derivative, the script maps\nthe host onto the upstream Ubuntu\u002FDebian CUDA repo via `UBUNTU_CODENAME` \u002F\n`VERSION_CODENAME` (`focal`\u002F`jammy`\u002F`noble` → Ubuntu 20.04\u002F22.04\u002F24.04;\n`bullseye`\u002F`bookworm`\u002F`trixie` → Debian 11\u002F12\u002F12). If the host's codename\ndoesn't match a known upstream release, `--install` exits with the same\nmanual-install guidance described above.\n\n**Docker not installed**: `--check-only` reports `MISSING: Docker is not\ninstalled` and prints the exact rerun command appropriate to the detected\ndistro family. The default `--install` path installs Docker (`docker.io` \u002F\n`moby-engine` \u002F `docker-ce` \u002F `docker` depending on family), enables\u002Fstarts\nthe daemon, configures the NVIDIA runtime, and adds the invoking user to\nthe `docker` group. If you prefer to manage Docker yourself, install it\nbefore rerunning the script or pass `--skip-docker-install`.\n\n**Docker installed but `docker run` still needs sudo**: The script adds the\ninvoking user to the `docker` group, but Linux only refreshes group\nmembership on a new login session. Log out and back in, or run\n`newgrp docker` in each new shell, until the new membership is active.\n\n**Docker runtime still missing**: Restart Docker, then rerun\n`nvidia-ctk runtime configure --runtime=docker`.\n\n**Driver branch detected != 580**: The driver-branch pin is exact on\ndebian-family (`nvidia-open-580`). On rhel-\u002Fsuse-family the script\ninstalls the latest open driver shipped in NVIDIA's CUDA 13.0 repo for\nthe detected distro, which is always ≥ 580. If your host needs a stricter\npin, set `$NVIDIA_DRIVER_PACKAGE_RHEL` \u002F `$NVIDIA_DRIVER_KMOD_RHEL` \u002F\n`$NVIDIA_DRIVER_PACKAGE_SUSE` to the exact package names you want before\nrunning `--install`.\n\n**Driver installed but `nvidia-smi` fails**: Load the module with\n`sudo modprobe nvidia` or reboot. Secure Boot may require MOK enrollment on\nsystems where it is enabled.\n\n**Kubernetes still has no GPU capacity**: Confirm the driver works on each GPU\nnode with `nvidia-smi`, then check the GPU Operator\u002Fdevice plugin pods and node\nlabels.\n",{"data":35,"body":44},{"name":4,"description":6,"license":26,"compatibility":36,"metadata":37,"allowed-tools":40,"tags":41},"Runs `--check-only` on any Linux distribution. `--install` automates Ubuntu 22.04\u002F24.04 + Debian 12 (apt), Fedora + RHEL\u002FRocky\u002FAlmaLinux 9\u002F10 (dnf), and openSUSE Leap \u002F SLES 15 (zypper). Requires sudo\u002Froot, internet access to NVIDIA package repositories (and download.docker.com on rhel-family), and an x86_64 or aarch64 (sbsa) host. Other distributions (Arch, Alpine, Gentoo, NixOS, …) get a clear error that names the version targets and the NVIDIA install-guide URL.",{"author":38,"version":39},"NVIDIA Corporation","0.1.0","Read Bash",[42,8,43,16,19],"setup","cuda",{"type":45,"children":46},"root",[47,56,85,191,228,241,497,504,509,648,702,708,713,916,921,927,939,944,1117,1122,1403,1409,1414,1535,1578,1584,1589,1610,1702,1707,1713,1738,1768,1780,1892,1961,1993,2009,2051,2076,2093],{"type":48,"tag":49,"props":50,"children":52},"element","h1",{"id":51},"nvidia-gpu-host-setup",[53],{"type":54,"value":55},"text","NVIDIA GPU Host Setup",{"type":48,"tag":57,"props":58,"children":59},"p",{},[60,62,68,70,76,78,83],{"type":54,"value":61},"Use this setup skill before TAO workflows run on the ",{"type":48,"tag":63,"props":64,"children":66},"code",{"className":65},[],[67],{"type":54,"value":16},{"type":54,"value":69},", ",{"type":48,"tag":63,"props":71,"children":73},{"className":72},[],[74],{"type":54,"value":75},"local-docker",{"type":54,"value":77},",\nor ",{"type":48,"tag":63,"props":79,"children":81},{"className":80},[],[82],{"type":54,"value":19},{"type":54,"value":84}," backend. It standardizes the host GPU runtime on:",{"type":48,"tag":86,"props":87,"children":88},"ul",{},[89,103,114,125],{"type":48,"tag":90,"props":91,"children":92},"li",{},[93,95,101],{"type":54,"value":94},"NVIDIA driver branch ",{"type":48,"tag":63,"props":96,"children":98},{"className":97},[],[99],{"type":54,"value":100},"580",{"type":54,"value":102}," (open kernel module preferred)",{"type":48,"tag":90,"props":104,"children":105},{},[106,108],{"type":54,"value":107},"CUDA Toolkit package ",{"type":48,"tag":63,"props":109,"children":111},{"className":110},[],[112],{"type":54,"value":113},"cuda-toolkit-13-0",{"type":48,"tag":90,"props":115,"children":116},{},[117,119],{"type":54,"value":118},"NVIDIA Container Toolkit ",{"type":48,"tag":63,"props":120,"children":122},{"className":121},[],[123],{"type":54,"value":124},"1.19.0",{"type":48,"tag":90,"props":126,"children":127},{},[128,130,135,137,142,144,150,152,158,160,166,168,174,176,181,183,189],{"type":54,"value":129},"Docker engine — only installed for ",{"type":48,"tag":63,"props":131,"children":133},{"className":132},[],[134],{"type":54,"value":16},{"type":54,"value":136}," \u002F ",{"type":48,"tag":63,"props":138,"children":140},{"className":139},[],[141],{"type":54,"value":75},{"type":54,"value":143}," backends and\nonly when Docker is missing. The package picked depends on the distro\nfamily (",{"type":48,"tag":63,"props":145,"children":147},{"className":146},[],[148],{"type":54,"value":149},"docker.io",{"type":54,"value":151}," on Debian-family by default, ",{"type":48,"tag":63,"props":153,"children":155},{"className":154},[],[156],{"type":54,"value":157},"moby-engine",{"type":54,"value":159}," \u002F\n",{"type":48,"tag":63,"props":161,"children":163},{"className":162},[],[164],{"type":54,"value":165},"docker-ce",{"type":54,"value":167}," from ",{"type":48,"tag":63,"props":169,"children":171},{"className":170},[],[172],{"type":54,"value":173},"download.docker.com",{"type":54,"value":175}," on RHEL-family, ",{"type":48,"tag":63,"props":177,"children":179},{"className":178},[],[180],{"type":54,"value":16},{"type":54,"value":182}," on\nSUSE-family). Pass ",{"type":48,"tag":63,"props":184,"children":186},{"className":185},[],[187],{"type":54,"value":188},"--skip-docker-install",{"type":54,"value":190}," to opt out.",{"type":48,"tag":57,"props":192,"children":193},{},[194,196,202,204,210,212,218,220,226],{"type":54,"value":195},"The check is safe and read-only by default — it works on any Linux\ndistribution because it only probes ",{"type":48,"tag":63,"props":197,"children":199},{"className":198},[],[200],{"type":54,"value":201},"nvidia-smi",{"type":54,"value":203},", the CUDA toolkit path,\nthe installed container-toolkit package version (via ",{"type":48,"tag":63,"props":205,"children":207},{"className":206},[],[208],{"type":54,"value":209},"dpkg",{"type":54,"value":211},"\u002F",{"type":48,"tag":63,"props":213,"children":215},{"className":214},[],[216],{"type":54,"value":217},"rpm",{"type":54,"value":219},"\u002Fthe\n",{"type":48,"tag":63,"props":221,"children":223},{"className":222},[],[224],{"type":54,"value":225},"nvidia-ctk",{"type":54,"value":227}," binary version), and the Docker daemon's NVIDIA runtime.",{"type":48,"tag":57,"props":229,"children":230},{},[231,233,239],{"type":54,"value":232},"Installation must be explicitly authorized by the user and rerun with\n",{"type":48,"tag":63,"props":234,"children":236},{"className":235},[],[237],{"type":54,"value":238},"--install",{"type":54,"value":240},". The install path is automated for these distro families:",{"type":48,"tag":242,"props":243,"children":244},"table",{},[245,274],{"type":48,"tag":246,"props":247,"children":248},"thead",{},[249],{"type":48,"tag":250,"props":251,"children":252},"tr",{},[253,259,264,269],{"type":48,"tag":254,"props":255,"children":256},"th",{},[257],{"type":54,"value":258},"Family",{"type":48,"tag":254,"props":260,"children":261},{},[262],{"type":54,"value":263},"Tested distros",{"type":48,"tag":254,"props":265,"children":266},{},[267],{"type":54,"value":268},"Manager",{"type":48,"tag":254,"props":270,"children":271},{},[272],{"type":54,"value":273},"Notes",{"type":48,"tag":275,"props":276,"children":277},"tbody",{},[278,352,422,463],{"type":48,"tag":250,"props":279,"children":280},{},[281,287,307,316],{"type":48,"tag":282,"props":283,"children":284},"td",{},[285],{"type":54,"value":286},"debian",{"type":48,"tag":282,"props":288,"children":289},{},[290,292,298,299,305],{"type":54,"value":291},"Ubuntu 22.04 \u002F 24.04, Debian 12 (and derivatives Pop!_OS, Mint, Zorin, Raspbian, KDE Neon, etc. via ",{"type":48,"tag":63,"props":293,"children":295},{"className":294},[],[296],{"type":54,"value":297},"UBUNTU_CODENAME",{"type":54,"value":136},{"type":48,"tag":63,"props":300,"children":302},{"className":301},[],[303],{"type":54,"value":304},"VERSION_CODENAME",{"type":54,"value":306},")",{"type":48,"tag":282,"props":308,"children":309},{},[310],{"type":48,"tag":63,"props":311,"children":313},{"className":312},[],[314],{"type":54,"value":315},"apt-get",{"type":48,"tag":282,"props":317,"children":318},{},[319,321,327,329,335,337,342,344,350],{"type":54,"value":320},"Adds NVIDIA ",{"type":48,"tag":63,"props":322,"children":324},{"className":323},[],[325],{"type":54,"value":326},"cuda-keyring",{"type":54,"value":328}," + Container Toolkit ",{"type":48,"tag":63,"props":330,"children":332},{"className":331},[],[333],{"type":54,"value":334},".list",{"type":54,"value":336},". Docker via ",{"type":48,"tag":63,"props":338,"children":340},{"className":339},[],[341],{"type":54,"value":149},{"type":54,"value":343}," (override ",{"type":48,"tag":63,"props":345,"children":347},{"className":346},[],[348],{"type":54,"value":349},"$DOCKER_PACKAGE_DEBIAN",{"type":54,"value":351},").",{"type":48,"tag":250,"props":353,"children":354},{},[355,360,365,383],{"type":48,"tag":282,"props":356,"children":357},{},[358],{"type":54,"value":359},"rhel",{"type":48,"tag":282,"props":361,"children":362},{},[363],{"type":54,"value":364},"Fedora 39+, RHEL \u002F Rocky \u002F AlmaLinux 9 and 10",{"type":48,"tag":282,"props":366,"children":367},{},[368,374,376,382],{"type":48,"tag":63,"props":369,"children":371},{"className":370},[],[372],{"type":54,"value":373},"dnf",{"type":54,"value":375}," (or ",{"type":48,"tag":63,"props":377,"children":379},{"className":378},[],[380],{"type":54,"value":381},"yum",{"type":54,"value":306},{"type":48,"tag":282,"props":384,"children":385},{},[386,387,393,394,400,402,407,409,414,415,420],{"type":54,"value":320},{"type":48,"tag":63,"props":388,"children":390},{"className":389},[],[391],{"type":54,"value":392},"cuda-\u003Cdistro>.repo",{"type":54,"value":328},{"type":48,"tag":63,"props":395,"children":397},{"className":396},[],[398],{"type":54,"value":399},".repo",{"type":54,"value":401},". Docker via Fedora ",{"type":48,"tag":63,"props":403,"children":405},{"className":404},[],[406],{"type":54,"value":157},{"type":54,"value":408}," when available, otherwise ",{"type":48,"tag":63,"props":410,"children":412},{"className":411},[],[413],{"type":54,"value":165},{"type":54,"value":167},{"type":48,"tag":63,"props":416,"children":418},{"className":417},[],[419],{"type":54,"value":173},{"type":54,"value":421},".",{"type":48,"tag":250,"props":423,"children":424},{},[425,430,435,444],{"type":48,"tag":282,"props":426,"children":427},{},[428],{"type":54,"value":429},"suse",{"type":48,"tag":282,"props":431,"children":432},{},[433],{"type":54,"value":434},"openSUSE Leap 15, SLES 15",{"type":48,"tag":282,"props":436,"children":437},{},[438],{"type":48,"tag":63,"props":439,"children":441},{"className":440},[],[442],{"type":54,"value":443},"zypper",{"type":48,"tag":282,"props":445,"children":446},{},[447,449,454,456,461],{"type":54,"value":448},"Adds the same NVIDIA ",{"type":48,"tag":63,"props":450,"children":452},{"className":451},[],[453],{"type":54,"value":399},{"type":54,"value":455}," files. Docker via the distribution ",{"type":48,"tag":63,"props":457,"children":459},{"className":458},[],[460],{"type":54,"value":16},{"type":54,"value":462}," package.",{"type":48,"tag":250,"props":464,"children":465},{},[466,471,476,480],{"type":48,"tag":282,"props":467,"children":468},{},[469],{"type":54,"value":470},"other (Arch, Alpine, Gentoo, NixOS, FreeBSD, …)",{"type":48,"tag":282,"props":472,"children":473},{},[474],{"type":54,"value":475},"n\u002Fa",{"type":48,"tag":282,"props":477,"children":478},{},[479],{"type":54,"value":475},{"type":48,"tag":282,"props":481,"children":482},{},[483,488,490,496],{"type":48,"tag":63,"props":484,"children":486},{"className":485},[],[487],{"type":54,"value":238},{"type":54,"value":489}," exits with a clear error listing the version targets and the NVIDIA install-guide URLs. Install manually, then rerun ",{"type":48,"tag":63,"props":491,"children":493},{"className":492},[],[494],{"type":54,"value":495},"--check-only",{"type":54,"value":421},{"type":48,"tag":498,"props":499,"children":501},"h2",{"id":500},"quick-start",[502],{"type":54,"value":503},"Quick Start",{"type":48,"tag":57,"props":505,"children":506},{},[507],{"type":54,"value":508},"From the skill bank root:",{"type":48,"tag":510,"props":511,"children":516},"pre",{"className":512,"code":513,"language":514,"meta":515,"style":515},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","# Check the local Docker backend host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n\n# Install or repair after user approval.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --install\n\n# Check a Kubernetes GPU worker host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend kubernetes --check-only\n","bash","",[517],{"type":48,"tag":63,"props":518,"children":519},{"__ignoreMap":515},[520,532,562,572,581,606,614,623],{"type":48,"tag":521,"props":522,"children":525},"span",{"class":523,"line":524},"line",1,[526],{"type":48,"tag":521,"props":527,"children":529},{"style":528},"--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic",[530],{"type":54,"value":531},"# Check the local Docker backend host.\n",{"type":48,"tag":521,"props":533,"children":535},{"class":523,"line":534},2,[536,541,547,552,557],{"type":48,"tag":521,"props":537,"children":539},{"style":538},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[540],{"type":54,"value":514},{"type":48,"tag":521,"props":542,"children":544},{"style":543},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[545],{"type":54,"value":546}," skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh",{"type":48,"tag":521,"props":548,"children":549},{"style":543},[550],{"type":54,"value":551}," --backend",{"type":48,"tag":521,"props":553,"children":554},{"style":543},[555],{"type":54,"value":556}," docker",{"type":48,"tag":521,"props":558,"children":559},{"style":543},[560],{"type":54,"value":561}," --check-only\n",{"type":48,"tag":521,"props":563,"children":565},{"class":523,"line":564},3,[566],{"type":48,"tag":521,"props":567,"children":569},{"emptyLinePlaceholder":568},true,[570],{"type":54,"value":571},"\n",{"type":48,"tag":521,"props":573,"children":575},{"class":523,"line":574},4,[576],{"type":48,"tag":521,"props":577,"children":578},{"style":528},[579],{"type":54,"value":580},"# Install or repair after user approval.\n",{"type":48,"tag":521,"props":582,"children":584},{"class":523,"line":583},5,[585,589,593,597,601],{"type":48,"tag":521,"props":586,"children":587},{"style":538},[588],{"type":54,"value":514},{"type":48,"tag":521,"props":590,"children":591},{"style":543},[592],{"type":54,"value":546},{"type":48,"tag":521,"props":594,"children":595},{"style":543},[596],{"type":54,"value":551},{"type":48,"tag":521,"props":598,"children":599},{"style":543},[600],{"type":54,"value":556},{"type":48,"tag":521,"props":602,"children":603},{"style":543},[604],{"type":54,"value":605}," --install\n",{"type":48,"tag":521,"props":607,"children":609},{"class":523,"line":608},6,[610],{"type":48,"tag":521,"props":611,"children":612},{"emptyLinePlaceholder":568},[613],{"type":54,"value":571},{"type":48,"tag":521,"props":615,"children":617},{"class":523,"line":616},7,[618],{"type":48,"tag":521,"props":619,"children":620},{"style":528},[621],{"type":54,"value":622},"# Check a Kubernetes GPU worker host.\n",{"type":48,"tag":521,"props":624,"children":626},{"class":523,"line":625},8,[627,631,635,639,644],{"type":48,"tag":521,"props":628,"children":629},{"style":538},[630],{"type":54,"value":514},{"type":48,"tag":521,"props":632,"children":633},{"style":543},[634],{"type":54,"value":546},{"type":48,"tag":521,"props":636,"children":637},{"style":543},[638],{"type":54,"value":551},{"type":48,"tag":521,"props":640,"children":641},{"style":543},[642],{"type":54,"value":643}," kubernetes",{"type":48,"tag":521,"props":645,"children":646},{"style":543},[647],{"type":54,"value":561},{"type":48,"tag":649,"props":650,"children":651},"blockquote",{},[652],{"type":48,"tag":57,"props":653,"children":654},{},[655,657,663,665,671,673,678,680,686,688,693,695,700],{"type":54,"value":656},"⚠️ ",{"type":48,"tag":658,"props":659,"children":660},"strong",{},[661],{"type":54,"value":662},"Note — running non-interactively (agent\u002Fskill runs):",{"type":54,"value":664}," a skill run has no terminal, so the\ninstaller's ",{"type":48,"tag":63,"props":666,"children":668},{"className":667},[],[669],{"type":54,"value":670},"Continue? [y\u002FN]",{"type":54,"value":672}," prompt cannot be answered. After running ",{"type":48,"tag":63,"props":674,"children":676},{"className":675},[],[677],{"type":54,"value":495},{"type":54,"value":679}," to preview and\ngetting the user's approval, append the assume-yes flag (",{"type":48,"tag":63,"props":681,"children":683},{"className":682},[],[684],{"type":54,"value":685},"--yes",{"type":54,"value":687},") to the ",{"type":48,"tag":63,"props":689,"children":691},{"className":690},[],[692],{"type":54,"value":238},{"type":54,"value":694}," command so it\nproceeds without a prompt — this auto-confirms installation of system packages (NVIDIA driver, CUDA\nToolkit, NVIDIA Container Toolkit, and Docker for Docker backends) and modifies the host, so only do\nthis on a host you control. A person running ",{"type":48,"tag":63,"props":696,"children":698},{"className":697},[],[699],{"type":54,"value":238},{"type":54,"value":701}," directly at a terminal gets the prompt instead.",{"type":48,"tag":498,"props":703,"children":705},{"id":704},"workflow-contract",[706],{"type":54,"value":707},"Workflow Contract",{"type":48,"tag":57,"props":709,"children":710},{},[711],{"type":54,"value":712},"Docker and Kubernetes workflows must run the check before submitting GPU work:",{"type":48,"tag":510,"props":714,"children":716},{"className":512,"code":715,"language":514,"meta":515,"style":515},"SETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n\nbash \"$SETUP_SCRIPT\" --backend docker --check-only || {\n  echo \"MISSING: TAO GPU host runtime is not ready.\"\n  echo \"After user approval, run (append --yes for non-interactive agent runs):\"\n  echo \"  bash \\\"$SETUP_SCRIPT\\\" --backend docker --install\"\n  exit 1\n}\n",[717],{"type":48,"tag":63,"props":718,"children":719},{"__ignoreMap":515},[720,770,777,822,844,864,894,908],{"type":48,"tag":521,"props":721,"children":722},{"class":523,"line":524},[723,729,735,740,745,750,755,760,765],{"type":48,"tag":521,"props":724,"children":726},{"style":725},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[727],{"type":54,"value":728},"SETUP_SCRIPT",{"type":48,"tag":521,"props":730,"children":732},{"style":731},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[733],{"type":54,"value":734},"=",{"type":48,"tag":521,"props":736,"children":737},{"style":731},[738],{"type":54,"value":739},"\"${",{"type":48,"tag":521,"props":741,"children":742},{"style":725},[743],{"type":54,"value":744},"TAO_SKILL_BANK_ROOT",{"type":48,"tag":521,"props":746,"children":747},{"style":731},[748],{"type":54,"value":749},":-",{"type":48,"tag":521,"props":751,"children":752},{"style":725},[753],{"type":54,"value":754},"$PWD",{"type":48,"tag":521,"props":756,"children":757},{"style":731},[758],{"type":54,"value":759},"}",{"type":48,"tag":521,"props":761,"children":762},{"style":543},[763],{"type":54,"value":764},"\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh",{"type":48,"tag":521,"props":766,"children":767},{"style":731},[768],{"type":54,"value":769},"\"\n",{"type":48,"tag":521,"props":771,"children":772},{"class":523,"line":534},[773],{"type":48,"tag":521,"props":774,"children":775},{"emptyLinePlaceholder":568},[776],{"type":54,"value":571},{"type":48,"tag":521,"props":778,"children":779},{"class":523,"line":564},[780,784,789,794,799,803,807,812,817],{"type":48,"tag":521,"props":781,"children":782},{"style":538},[783],{"type":54,"value":514},{"type":48,"tag":521,"props":785,"children":786},{"style":731},[787],{"type":54,"value":788}," \"",{"type":48,"tag":521,"props":790,"children":791},{"style":725},[792],{"type":54,"value":793},"$SETUP_SCRIPT",{"type":48,"tag":521,"props":795,"children":796},{"style":731},[797],{"type":54,"value":798},"\"",{"type":48,"tag":521,"props":800,"children":801},{"style":543},[802],{"type":54,"value":551},{"type":48,"tag":521,"props":804,"children":805},{"style":543},[806],{"type":54,"value":556},{"type":48,"tag":521,"props":808,"children":809},{"style":543},[810],{"type":54,"value":811}," --check-only",{"type":48,"tag":521,"props":813,"children":814},{"style":731},[815],{"type":54,"value":816}," ||",{"type":48,"tag":521,"props":818,"children":819},{"style":731},[820],{"type":54,"value":821}," {\n",{"type":48,"tag":521,"props":823,"children":824},{"class":523,"line":574},[825,831,835,840],{"type":48,"tag":521,"props":826,"children":828},{"style":827},"--shiki-light:#6182B8;--shiki-default:#82AAFF;--shiki-dark:#82AAFF",[829],{"type":54,"value":830},"  echo",{"type":48,"tag":521,"props":832,"children":833},{"style":731},[834],{"type":54,"value":788},{"type":48,"tag":521,"props":836,"children":837},{"style":543},[838],{"type":54,"value":839},"MISSING: TAO GPU host runtime is not ready.",{"type":48,"tag":521,"props":841,"children":842},{"style":731},[843],{"type":54,"value":769},{"type":48,"tag":521,"props":845,"children":846},{"class":523,"line":583},[847,851,855,860],{"type":48,"tag":521,"props":848,"children":849},{"style":827},[850],{"type":54,"value":830},{"type":48,"tag":521,"props":852,"children":853},{"style":731},[854],{"type":54,"value":788},{"type":48,"tag":521,"props":856,"children":857},{"style":543},[858],{"type":54,"value":859},"After user approval, run (append --yes for non-interactive agent runs):",{"type":48,"tag":521,"props":861,"children":862},{"style":731},[863],{"type":54,"value":769},{"type":48,"tag":521,"props":865,"children":866},{"class":523,"line":608},[867,871,875,880,885,890],{"type":48,"tag":521,"props":868,"children":869},{"style":827},[870],{"type":54,"value":830},{"type":48,"tag":521,"props":872,"children":873},{"style":731},[874],{"type":54,"value":788},{"type":48,"tag":521,"props":876,"children":877},{"style":543},[878],{"type":54,"value":879},"  bash ",{"type":48,"tag":521,"props":881,"children":882},{"style":725},[883],{"type":54,"value":884},"\\\"$SETUP_SCRIPT\\\"",{"type":48,"tag":521,"props":886,"children":887},{"style":543},[888],{"type":54,"value":889}," --backend docker --install",{"type":48,"tag":521,"props":891,"children":892},{"style":731},[893],{"type":54,"value":769},{"type":48,"tag":521,"props":895,"children":896},{"class":523,"line":616},[897,902],{"type":48,"tag":521,"props":898,"children":899},{"style":827},[900],{"type":54,"value":901},"  exit",{"type":48,"tag":521,"props":903,"children":905},{"style":904},"--shiki-light:#F76D47;--shiki-default:#F78C6C;--shiki-dark:#F78C6C",[906],{"type":54,"value":907}," 1\n",{"type":48,"tag":521,"props":909,"children":910},{"class":523,"line":625},[911],{"type":48,"tag":521,"props":912,"children":913},{"style":731},[914],{"type":54,"value":915},"}\n",{"type":48,"tag":57,"props":917,"children":918},{},[919],{"type":54,"value":920},"Never install silently. If the check fails, explain what is missing, ask the\nuser to authorize the fix, then run the install command and rerun the check.",{"type":48,"tag":498,"props":922,"children":924},{"id":923},"what-the-installer-does",[925],{"type":54,"value":926},"What The Installer Does",{"type":48,"tag":57,"props":928,"children":929},{},[930,932,937],{"type":54,"value":931},"The installer dispatches on the detected distribution family. On every\nsupported family it adds NVIDIA's CUDA and Container Toolkit repositories\n(if missing), installs the pinned runtime packages, optionally installs\nDocker, wires the NVIDIA Docker runtime, and adds the invoking user to\nthe ",{"type":48,"tag":63,"props":933,"children":935},{"className":934},[],[936],{"type":54,"value":16},{"type":54,"value":938}," group.",{"type":48,"tag":57,"props":940,"children":941},{},[942],{"type":54,"value":943},"Common steps (all families):",{"type":48,"tag":945,"props":946,"children":947},"ol",{},[948,967,985,990,1017,1038,1104],{"type":48,"tag":90,"props":949,"children":950},{},[951,953,958,960,965],{"type":54,"value":952},"Adds NVIDIA's CUDA repository if missing (apt ",{"type":48,"tag":63,"props":954,"children":956},{"className":955},[],[957],{"type":54,"value":326},{"type":54,"value":959}," deb,\n",{"type":48,"tag":63,"props":961,"children":963},{"className":962},[],[964],{"type":54,"value":392},{"type":54,"value":966}," for dnf\u002Fzypper).",{"type":48,"tag":90,"props":968,"children":969},{},[970,972,977,979,984],{"type":54,"value":971},"Adds NVIDIA's Container Toolkit repository if missing (",{"type":48,"tag":63,"props":973,"children":975},{"className":974},[],[976],{"type":54,"value":334},{"type":54,"value":978}," for apt,\n",{"type":48,"tag":63,"props":980,"children":982},{"className":981},[],[983],{"type":54,"value":399},{"type":54,"value":966},{"type":48,"tag":90,"props":986,"children":987},{},[988],{"type":54,"value":989},"Installs the matching kernel header \u002F devel package for the running\nkernel.",{"type":48,"tag":90,"props":991,"children":992},{},[993,995,1000,1002,1007,1009,1015],{"type":54,"value":994},"Installs the driver branch 580 packages, ",{"type":48,"tag":63,"props":996,"children":998},{"className":997},[],[999],{"type":54,"value":113},{"type":54,"value":1001},", and the\nContainer Toolkit pinned to ",{"type":48,"tag":63,"props":1003,"children":1005},{"className":1004},[],[1006],{"type":54,"value":124},{"type":54,"value":1008}," (the dpkg-suffixed ",{"type":48,"tag":63,"props":1010,"children":1012},{"className":1011},[],[1013],{"type":54,"value":1014},"1.19.0-1",{"type":54,"value":1016}," is\nthe same upstream version expressed for apt).",{"type":48,"tag":90,"props":1018,"children":1019},{},[1020,1022,1028,1030,1036],{"type":54,"value":1021},"For Docker backends and when Docker is missing, installs Docker\n(override \u002F opt-out flags below), enables\u002Fstarts the daemon, then runs\n",{"type":48,"tag":63,"props":1023,"children":1025},{"className":1024},[],[1026],{"type":54,"value":1027},"nvidia-ctk runtime configure --runtime=docker",{"type":54,"value":1029}," and restarts Docker\nwhen ",{"type":48,"tag":63,"props":1031,"children":1033},{"className":1032},[],[1034],{"type":54,"value":1035},"systemctl",{"type":54,"value":1037}," is available.",{"type":48,"tag":90,"props":1039,"children":1040},{},[1041,1043,1049,1051,1057,1059,1064,1066,1071,1073,1079,1081,1087,1089,1094,1096,1102],{"type":54,"value":1042},"Adds the invoking user (",{"type":48,"tag":63,"props":1044,"children":1046},{"className":1045},[],[1047],{"type":54,"value":1048},"$SUDO_USER",{"type":54,"value":1050}," if available, else ",{"type":48,"tag":63,"props":1052,"children":1054},{"className":1053},[],[1055],{"type":54,"value":1056},"$USER",{"type":54,"value":1058},") to the\n",{"type":48,"tag":63,"props":1060,"children":1062},{"className":1061},[],[1063],{"type":54,"value":16},{"type":54,"value":1065}," group so subsequent shells can run ",{"type":48,"tag":63,"props":1067,"children":1069},{"className":1068},[],[1070],{"type":54,"value":16},{"type":54,"value":1072}," without ",{"type":48,"tag":63,"props":1074,"children":1076},{"className":1075},[],[1077],{"type":54,"value":1078},"sudo",{"type":54,"value":1080}," —\nopt out with ",{"type":48,"tag":63,"props":1082,"children":1084},{"className":1083},[],[1085],{"type":54,"value":1086},"--skip-docker-group",{"type":54,"value":1088},". ",{"type":48,"tag":658,"props":1090,"children":1091},{},[1092],{"type":54,"value":1093},"The new group membership does not\ntake effect in the current shell",{"type":54,"value":1095},": log out and back in, or run\n",{"type":48,"tag":63,"props":1097,"children":1099},{"className":1098},[],[1100],{"type":54,"value":1101},"newgrp docker",{"type":54,"value":1103}," in each new shell.",{"type":48,"tag":90,"props":1105,"children":1106},{},[1107,1109,1115],{"type":54,"value":1108},"Attempts ",{"type":48,"tag":63,"props":1110,"children":1112},{"className":1111},[],[1113],{"type":54,"value":1114},"modprobe nvidia",{"type":54,"value":1116}," so verification can pass before reboot.",{"type":48,"tag":57,"props":1118,"children":1119},{},[1120],{"type":54,"value":1121},"Family-specific package selections:",{"type":48,"tag":242,"props":1123,"children":1124},{},[1125,1151],{"type":48,"tag":246,"props":1126,"children":1127},{},[1128],{"type":48,"tag":250,"props":1129,"children":1130},{},[1131,1136,1141,1146],{"type":48,"tag":254,"props":1132,"children":1133},{},[1134],{"type":54,"value":1135},"Step",{"type":48,"tag":254,"props":1137,"children":1138},{},[1139],{"type":54,"value":1140},"debian-family",{"type":48,"tag":254,"props":1142,"children":1143},{},[1144],{"type":54,"value":1145},"rhel-family",{"type":48,"tag":254,"props":1147,"children":1148},{},[1149],{"type":54,"value":1150},"suse-family",{"type":48,"tag":275,"props":1152,"children":1153},{},[1154,1196,1277,1309,1343],{"type":48,"tag":250,"props":1155,"children":1156},{},[1157,1162,1171,1187],{"type":48,"tag":282,"props":1158,"children":1159},{},[1160],{"type":54,"value":1161},"Kernel headers",{"type":48,"tag":282,"props":1163,"children":1164},{},[1165],{"type":48,"tag":63,"props":1166,"children":1168},{"className":1167},[],[1169],{"type":54,"value":1170},"linux-headers-$(uname -r)",{"type":48,"tag":282,"props":1172,"children":1173},{},[1174,1180,1181],{"type":48,"tag":63,"props":1175,"children":1177},{"className":1176},[],[1178],{"type":54,"value":1179},"kernel-devel-$(uname -r)",{"type":54,"value":69},{"type":48,"tag":63,"props":1182,"children":1184},{"className":1183},[],[1185],{"type":54,"value":1186},"kernel-headers-$(uname -r)",{"type":48,"tag":282,"props":1188,"children":1189},{},[1190],{"type":48,"tag":63,"props":1191,"children":1193},{"className":1192},[],[1194],{"type":54,"value":1195},"kernel-default-devel",{"type":48,"tag":250,"props":1197,"children":1198},{},[1199,1204,1229,1260],{"type":48,"tag":282,"props":1200,"children":1201},{},[1202],{"type":54,"value":1203},"Driver",{"type":48,"tag":282,"props":1205,"children":1206},{},[1207,1213,1214,1220,1222,1228],{"type":48,"tag":63,"props":1208,"children":1210},{"className":1209},[],[1211],{"type":54,"value":1212},"nvidia-driver-pinning-580",{"type":54,"value":69},{"type":48,"tag":63,"props":1215,"children":1217},{"className":1216},[],[1218],{"type":54,"value":1219},"nvidia-open-580",{"type":54,"value":1221}," (override: ",{"type":48,"tag":63,"props":1223,"children":1225},{"className":1224},[],[1226],{"type":54,"value":1227},"$NVIDIA_DRIVER_PACKAGE_DEBIAN",{"type":54,"value":306},{"type":48,"tag":282,"props":1230,"children":1231},{},[1232,1238,1239,1245,1246,1252,1253,1259],{"type":48,"tag":63,"props":1233,"children":1235},{"className":1234},[],[1236],{"type":54,"value":1237},"nvidia-driver-cuda",{"type":54,"value":69},{"type":48,"tag":63,"props":1240,"children":1242},{"className":1241},[],[1243],{"type":54,"value":1244},"kmod-nvidia-open-dkms",{"type":54,"value":1221},{"type":48,"tag":63,"props":1247,"children":1249},{"className":1248},[],[1250],{"type":54,"value":1251},"$NVIDIA_DRIVER_PACKAGE_RHEL",{"type":54,"value":69},{"type":48,"tag":63,"props":1254,"children":1256},{"className":1255},[],[1257],{"type":54,"value":1258},"$NVIDIA_DRIVER_KMOD_RHEL",{"type":54,"value":306},{"type":48,"tag":282,"props":1261,"children":1262},{},[1263,1269,1270,1276],{"type":48,"tag":63,"props":1264,"children":1266},{"className":1265},[],[1267],{"type":54,"value":1268},"nvidia-open-driver-G06-signed-kmp-default",{"type":54,"value":1221},{"type":48,"tag":63,"props":1271,"children":1273},{"className":1272},[],[1274],{"type":54,"value":1275},"$NVIDIA_DRIVER_PACKAGE_SUSE",{"type":54,"value":306},{"type":48,"tag":250,"props":1278,"children":1279},{},[1280,1285,1293,1301],{"type":48,"tag":282,"props":1281,"children":1282},{},[1283],{"type":54,"value":1284},"CUDA toolkit",{"type":48,"tag":282,"props":1286,"children":1287},{},[1288],{"type":48,"tag":63,"props":1289,"children":1291},{"className":1290},[],[1292],{"type":54,"value":113},{"type":48,"tag":282,"props":1294,"children":1295},{},[1296],{"type":48,"tag":63,"props":1297,"children":1299},{"className":1298},[],[1300],{"type":54,"value":113},{"type":48,"tag":282,"props":1302,"children":1303},{},[1304],{"type":48,"tag":63,"props":1305,"children":1307},{"className":1306},[],[1308],{"type":54,"value":113},{"type":48,"tag":250,"props":1310,"children":1311},{},[1312,1317,1328,1338],{"type":48,"tag":282,"props":1313,"children":1314},{},[1315],{"type":54,"value":1316},"Container Toolkit",{"type":48,"tag":282,"props":1318,"children":1319},{},[1320,1326],{"type":48,"tag":63,"props":1321,"children":1323},{"className":1322},[],[1324],{"type":54,"value":1325},"nvidia-container-toolkit=1.19.0-1",{"type":54,"value":1327}," + base\u002Ftools\u002Flibs",{"type":48,"tag":282,"props":1329,"children":1330},{},[1331,1337],{"type":48,"tag":63,"props":1332,"children":1334},{"className":1333},[],[1335],{"type":54,"value":1336},"nvidia-container-toolkit-1.19.0",{"type":54,"value":1327},{"type":48,"tag":282,"props":1339,"children":1340},{},[1341],{"type":54,"value":1342},"same as rhel",{"type":48,"tag":250,"props":1344,"children":1345},{},[1346,1350,1365,1395],{"type":48,"tag":282,"props":1347,"children":1348},{},[1349],{"type":54,"value":15},{"type":48,"tag":282,"props":1351,"children":1352},{},[1353,1358,1359,1364],{"type":48,"tag":63,"props":1354,"children":1356},{"className":1355},[],[1357],{"type":54,"value":149},{"type":54,"value":1221},{"type":48,"tag":63,"props":1360,"children":1362},{"className":1361},[],[1363],{"type":54,"value":349},{"type":54,"value":306},{"type":48,"tag":282,"props":1366,"children":1367},{},[1368,1373,1375,1381,1383,1389,1390],{"type":48,"tag":63,"props":1369,"children":1371},{"className":1370},[],[1372],{"type":54,"value":157},{"type":54,"value":1374},"+",{"type":48,"tag":63,"props":1376,"children":1378},{"className":1377},[],[1379],{"type":54,"value":1380},"moby-cli",{"type":54,"value":1382}," on Fedora when available, else ",{"type":48,"tag":63,"props":1384,"children":1386},{"className":1385},[],[1387],{"type":54,"value":1388},"docker-ce docker-ce-cli containerd.io",{"type":54,"value":167},{"type":48,"tag":63,"props":1391,"children":1393},{"className":1392},[],[1394],{"type":54,"value":173},{"type":48,"tag":282,"props":1396,"children":1397},{},[1398],{"type":48,"tag":63,"props":1399,"children":1401},{"className":1400},[],[1402],{"type":54,"value":16},{"type":48,"tag":498,"props":1404,"children":1406},{"id":1405},"verification",[1407],{"type":54,"value":1408},"Verification",{"type":48,"tag":57,"props":1410,"children":1411},{},[1412],{"type":54,"value":1413},"After installation, verify:",{"type":48,"tag":510,"props":1415,"children":1417},{"className":512,"code":1416,"language":514,"meta":515,"style":515},"nvidia-smi\n\u002Fusr\u002Flocal\u002Fcuda-13.0\u002Fbin\u002Fnvcc --version\ndocker info --format '{{json .Runtimes}}' | grep nvidia\nsudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smi\n",[1418],{"type":48,"tag":63,"props":1419,"children":1420},{"__ignoreMap":515},[1421,1429,1442,1489],{"type":48,"tag":521,"props":1422,"children":1423},{"class":523,"line":524},[1424],{"type":48,"tag":521,"props":1425,"children":1426},{"style":538},[1427],{"type":54,"value":1428},"nvidia-smi\n",{"type":48,"tag":521,"props":1430,"children":1431},{"class":523,"line":534},[1432,1437],{"type":48,"tag":521,"props":1433,"children":1434},{"style":538},[1435],{"type":54,"value":1436},"\u002Fusr\u002Flocal\u002Fcuda-13.0\u002Fbin\u002Fnvcc",{"type":48,"tag":521,"props":1438,"children":1439},{"style":543},[1440],{"type":54,"value":1441}," --version\n",{"type":48,"tag":521,"props":1443,"children":1444},{"class":523,"line":564},[1445,1449,1454,1459,1464,1469,1474,1479,1484],{"type":48,"tag":521,"props":1446,"children":1447},{"style":538},[1448],{"type":54,"value":16},{"type":48,"tag":521,"props":1450,"children":1451},{"style":543},[1452],{"type":54,"value":1453}," info",{"type":48,"tag":521,"props":1455,"children":1456},{"style":543},[1457],{"type":54,"value":1458}," --format",{"type":48,"tag":521,"props":1460,"children":1461},{"style":731},[1462],{"type":54,"value":1463}," '",{"type":48,"tag":521,"props":1465,"children":1466},{"style":543},[1467],{"type":54,"value":1468},"{{json .Runtimes}}",{"type":48,"tag":521,"props":1470,"children":1471},{"style":731},[1472],{"type":54,"value":1473},"'",{"type":48,"tag":521,"props":1475,"children":1476},{"style":731},[1477],{"type":54,"value":1478}," |",{"type":48,"tag":521,"props":1480,"children":1481},{"style":538},[1482],{"type":54,"value":1483}," grep",{"type":48,"tag":521,"props":1485,"children":1486},{"style":543},[1487],{"type":54,"value":1488}," nvidia\n",{"type":48,"tag":521,"props":1490,"children":1491},{"class":523,"line":574},[1492,1496,1500,1505,1510,1515,1520,1525,1530],{"type":48,"tag":521,"props":1493,"children":1494},{"style":538},[1495],{"type":54,"value":1078},{"type":48,"tag":521,"props":1497,"children":1498},{"style":543},[1499],{"type":54,"value":556},{"type":48,"tag":521,"props":1501,"children":1502},{"style":543},[1503],{"type":54,"value":1504}," run",{"type":48,"tag":521,"props":1506,"children":1507},{"style":543},[1508],{"type":54,"value":1509}," --rm",{"type":48,"tag":521,"props":1511,"children":1512},{"style":543},[1513],{"type":54,"value":1514}," --runtime=nvidia",{"type":48,"tag":521,"props":1516,"children":1517},{"style":543},[1518],{"type":54,"value":1519}," --gpus",{"type":48,"tag":521,"props":1521,"children":1522},{"style":543},[1523],{"type":54,"value":1524}," all",{"type":48,"tag":521,"props":1526,"children":1527},{"style":543},[1528],{"type":54,"value":1529}," ubuntu",{"type":48,"tag":521,"props":1531,"children":1532},{"style":543},[1533],{"type":54,"value":1534}," nvidia-smi\n",{"type":48,"tag":57,"props":1536,"children":1537},{},[1538,1540,1545,1547,1553,1555,1561,1563,1569,1571,1577],{"type":54,"value":1539},"Expected ",{"type":48,"tag":63,"props":1541,"children":1543},{"className":1542},[],[1544],{"type":54,"value":201},{"type":54,"value":1546}," output includes driver ",{"type":48,"tag":63,"props":1548,"children":1550},{"className":1549},[],[1551],{"type":54,"value":1552},"580.x",{"type":54,"value":1554}," and CUDA Version ",{"type":48,"tag":63,"props":1556,"children":1558},{"className":1557},[],[1559],{"type":54,"value":1560},"13.0",{"type":54,"value":1562},".\nExpected ",{"type":48,"tag":63,"props":1564,"children":1566},{"className":1565},[],[1567],{"type":54,"value":1568},"nvcc",{"type":54,"value":1570}," output includes ",{"type":48,"tag":63,"props":1572,"children":1574},{"className":1573},[],[1575],{"type":54,"value":1576},"release 13.0",{"type":54,"value":421},{"type":48,"tag":498,"props":1579,"children":1581},{"id":1580},"kubernetes-notes",[1582],{"type":54,"value":1583},"Kubernetes Notes",{"type":48,"tag":57,"props":1585,"children":1586},{},[1587],{"type":54,"value":1588},"For self-managed Kubernetes clusters, run the host installer on every GPU\nworker node or bake the same package set into the node image before installing\nthe NVIDIA GPU Operator or device plugin.",{"type":48,"tag":57,"props":1590,"children":1591},{},[1592,1594,1600,1602,1608],{"type":54,"value":1593},"The workflow check also warns if ",{"type":48,"tag":63,"props":1595,"children":1597},{"className":1596},[],[1598],{"type":54,"value":1599},"kubectl",{"type":54,"value":1601}," is available but the cluster reports\nno ",{"type":48,"tag":63,"props":1603,"children":1605},{"className":1604},[],[1606],{"type":54,"value":1607},"nvidia.com\u002Fgpu",{"type":54,"value":1609}," allocatable capacity. In that case, install\u002Fconfigure the\nNVIDIA GPU Operator after the worker host runtime is ready:",{"type":48,"tag":510,"props":1611,"children":1613},{"className":512,"code":1612,"language":514,"meta":515,"style":515},"helm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\nhelm install --wait gpu-operator -n gpu-operator --create-namespace nvidia\u002Fgpu-operator\n",[1614],{"type":48,"tag":63,"props":1615,"children":1616},{"__ignoreMap":515},[1617,1645,1661],{"type":48,"tag":521,"props":1618,"children":1619},{"class":523,"line":524},[1620,1625,1630,1635,1640],{"type":48,"tag":521,"props":1621,"children":1622},{"style":538},[1623],{"type":54,"value":1624},"helm",{"type":48,"tag":521,"props":1626,"children":1627},{"style":543},[1628],{"type":54,"value":1629}," repo",{"type":48,"tag":521,"props":1631,"children":1632},{"style":543},[1633],{"type":54,"value":1634}," add",{"type":48,"tag":521,"props":1636,"children":1637},{"style":543},[1638],{"type":54,"value":1639}," nvidia",{"type":48,"tag":521,"props":1641,"children":1642},{"style":543},[1643],{"type":54,"value":1644}," https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\n",{"type":48,"tag":521,"props":1646,"children":1647},{"class":523,"line":534},[1648,1652,1656],{"type":48,"tag":521,"props":1649,"children":1650},{"style":538},[1651],{"type":54,"value":1624},{"type":48,"tag":521,"props":1653,"children":1654},{"style":543},[1655],{"type":54,"value":1629},{"type":48,"tag":521,"props":1657,"children":1658},{"style":543},[1659],{"type":54,"value":1660}," update\n",{"type":48,"tag":521,"props":1662,"children":1663},{"class":523,"line":564},[1664,1668,1673,1678,1683,1688,1692,1697],{"type":48,"tag":521,"props":1665,"children":1666},{"style":538},[1667],{"type":54,"value":1624},{"type":48,"tag":521,"props":1669,"children":1670},{"style":543},[1671],{"type":54,"value":1672}," install",{"type":48,"tag":521,"props":1674,"children":1675},{"style":543},[1676],{"type":54,"value":1677}," --wait",{"type":48,"tag":521,"props":1679,"children":1680},{"style":543},[1681],{"type":54,"value":1682}," gpu-operator",{"type":48,"tag":521,"props":1684,"children":1685},{"style":543},[1686],{"type":54,"value":1687}," -n",{"type":48,"tag":521,"props":1689,"children":1690},{"style":543},[1691],{"type":54,"value":1682},{"type":48,"tag":521,"props":1693,"children":1694},{"style":543},[1695],{"type":54,"value":1696}," --create-namespace",{"type":48,"tag":521,"props":1698,"children":1699},{"style":543},[1700],{"type":54,"value":1701}," nvidia\u002Fgpu-operator\n",{"type":48,"tag":57,"props":1703,"children":1704},{},[1705],{"type":54,"value":1706},"Managed Kubernetes providers may own driver installation through node images or\nGPU Operator policy. Do not overwrite a provider-managed GPU node without user\napproval and a rollback plan.",{"type":48,"tag":498,"props":1708,"children":1710},{"id":1709},"failure-modes",[1711],{"type":54,"value":1712},"Failure Modes",{"type":48,"tag":57,"props":1714,"children":1715},{},[1716,1721,1723,1728,1730,1736],{"type":48,"tag":658,"props":1717,"children":1718},{},[1719],{"type":54,"value":1720},"Unsupported distribution family",{"type":54,"value":1722},": ",{"type":48,"tag":63,"props":1724,"children":1726},{"className":1725},[],[1727],{"type":54,"value":238},{"type":54,"value":1729}," automates debian-, rhel-,\nand suse-family hosts. On Arch, Alpine, Gentoo, NixOS, FreeBSD, or anything\nwithout ",{"type":48,"tag":63,"props":1731,"children":1733},{"className":1732},[],[1734],{"type":54,"value":1735},"\u002Fetc\u002Fos-release",{"type":54,"value":1737}," (e.g. macOS), the script exits with a clear error\nthat lists the four version targets and the upstream NVIDIA install-guide\nURLs:",{"type":48,"tag":86,"props":1739,"children":1740},{},[1741,1750,1759],{"type":48,"tag":90,"props":1742,"children":1743},{},[1744],{"type":48,"tag":63,"props":1745,"children":1747},{"className":1746},[],[1748],{"type":54,"value":1749},"https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fcuda-installation-guide-linux\u002F",{"type":48,"tag":90,"props":1751,"children":1752},{},[1753],{"type":48,"tag":63,"props":1754,"children":1756},{"className":1755},[],[1757],{"type":54,"value":1758},"https:\u002F\u002Fdocs.nvidia.com\u002Fdatacenter\u002Fcloud-native\u002Fcontainer-toolkit\u002Flatest\u002Finstall-guide.html",{"type":48,"tag":90,"props":1760,"children":1761},{},[1762],{"type":48,"tag":63,"props":1763,"children":1765},{"className":1764},[],[1766],{"type":54,"value":1767},"https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F",{"type":48,"tag":57,"props":1769,"children":1770},{},[1771,1773,1778],{"type":54,"value":1772},"Install those four pieces using your distribution's package manager and\nrerun the script with ",{"type":48,"tag":63,"props":1774,"children":1776},{"className":1775},[],[1777],{"type":54,"value":495},{"type":54,"value":1779}," to verify. The check is universally\nportable — it only queries the binaries \u002F package databases — so once the\nruntime is in place the workflow contract is satisfied regardless of the\nunderlying distro.",{"type":48,"tag":57,"props":1781,"children":1782},{},[1783,1788,1790,1796,1798,1804,1805,1811,1813,1819,1820,1826,1828,1833,1834,1839,1841,1847,1848,1854,1855,1861,1863,1869,1870,1876,1877,1883,1885,1890],{"type":48,"tag":658,"props":1784,"children":1785},{},[1786],{"type":54,"value":1787},"Unsupported Ubuntu\u002FDebian derivative",{"type":54,"value":1789},": When ",{"type":48,"tag":63,"props":1791,"children":1793},{"className":1792},[],[1794],{"type":54,"value":1795},"ID",{"type":54,"value":1797}," is e.g. ",{"type":48,"tag":63,"props":1799,"children":1801},{"className":1800},[],[1802],{"type":54,"value":1803},"pop",{"type":54,"value":69},{"type":48,"tag":63,"props":1806,"children":1808},{"className":1807},[],[1809],{"type":54,"value":1810},"mint",{"type":54,"value":1812},",\n",{"type":48,"tag":63,"props":1814,"children":1816},{"className":1815},[],[1817],{"type":54,"value":1818},"zorin",{"type":54,"value":69},{"type":48,"tag":63,"props":1821,"children":1823},{"className":1822},[],[1824],{"type":54,"value":1825},"raspbian",{"type":54,"value":1827},", or another debian-family derivative, the script maps\nthe host onto the upstream Ubuntu\u002FDebian CUDA repo via ",{"type":48,"tag":63,"props":1829,"children":1831},{"className":1830},[],[1832],{"type":54,"value":297},{"type":54,"value":159},{"type":48,"tag":63,"props":1835,"children":1837},{"className":1836},[],[1838],{"type":54,"value":304},{"type":54,"value":1840}," (",{"type":48,"tag":63,"props":1842,"children":1844},{"className":1843},[],[1845],{"type":54,"value":1846},"focal",{"type":54,"value":211},{"type":48,"tag":63,"props":1849,"children":1851},{"className":1850},[],[1852],{"type":54,"value":1853},"jammy",{"type":54,"value":211},{"type":48,"tag":63,"props":1856,"children":1858},{"className":1857},[],[1859],{"type":54,"value":1860},"noble",{"type":54,"value":1862}," → Ubuntu 20.04\u002F22.04\u002F24.04;\n",{"type":48,"tag":63,"props":1864,"children":1866},{"className":1865},[],[1867],{"type":54,"value":1868},"bullseye",{"type":54,"value":211},{"type":48,"tag":63,"props":1871,"children":1873},{"className":1872},[],[1874],{"type":54,"value":1875},"bookworm",{"type":54,"value":211},{"type":48,"tag":63,"props":1878,"children":1880},{"className":1879},[],[1881],{"type":54,"value":1882},"trixie",{"type":54,"value":1884}," → Debian 11\u002F12\u002F12). If the host's codename\ndoesn't match a known upstream release, ",{"type":48,"tag":63,"props":1886,"children":1888},{"className":1887},[],[1889],{"type":54,"value":238},{"type":54,"value":1891}," exits with the same\nmanual-install guidance described above.",{"type":48,"tag":57,"props":1893,"children":1894},{},[1895,1900,1901,1906,1908,1914,1916,1921,1923,1928,1929,1934,1935,1940,1941,1946,1948,1953,1955,1960],{"type":48,"tag":658,"props":1896,"children":1897},{},[1898],{"type":54,"value":1899},"Docker not installed",{"type":54,"value":1722},{"type":48,"tag":63,"props":1902,"children":1904},{"className":1903},[],[1905],{"type":54,"value":495},{"type":54,"value":1907}," reports ",{"type":48,"tag":63,"props":1909,"children":1911},{"className":1910},[],[1912],{"type":54,"value":1913},"MISSING: Docker is not installed",{"type":54,"value":1915}," and prints the exact rerun command appropriate to the detected\ndistro family. The default ",{"type":48,"tag":63,"props":1917,"children":1919},{"className":1918},[],[1920],{"type":54,"value":238},{"type":54,"value":1922}," path installs Docker (",{"type":48,"tag":63,"props":1924,"children":1926},{"className":1925},[],[1927],{"type":54,"value":149},{"type":54,"value":159},{"type":48,"tag":63,"props":1930,"children":1932},{"className":1931},[],[1933],{"type":54,"value":157},{"type":54,"value":136},{"type":48,"tag":63,"props":1936,"children":1938},{"className":1937},[],[1939],{"type":54,"value":165},{"type":54,"value":136},{"type":48,"tag":63,"props":1942,"children":1944},{"className":1943},[],[1945],{"type":54,"value":16},{"type":54,"value":1947}," depending on family), enables\u002Fstarts\nthe daemon, configures the NVIDIA runtime, and adds the invoking user to\nthe ",{"type":48,"tag":63,"props":1949,"children":1951},{"className":1950},[],[1952],{"type":54,"value":16},{"type":54,"value":1954}," group. If you prefer to manage Docker yourself, install it\nbefore rerunning the script or pass ",{"type":48,"tag":63,"props":1956,"children":1958},{"className":1957},[],[1959],{"type":54,"value":188},{"type":54,"value":421},{"type":48,"tag":57,"props":1962,"children":1963},{},[1964,1977,1979,1984,1986,1991],{"type":48,"tag":658,"props":1965,"children":1966},{},[1967,1969,1975],{"type":54,"value":1968},"Docker installed but ",{"type":48,"tag":63,"props":1970,"children":1972},{"className":1971},[],[1973],{"type":54,"value":1974},"docker run",{"type":54,"value":1976}," still needs sudo",{"type":54,"value":1978},": The script adds the\ninvoking user to the ",{"type":48,"tag":63,"props":1980,"children":1982},{"className":1981},[],[1983],{"type":54,"value":16},{"type":54,"value":1985}," group, but Linux only refreshes group\nmembership on a new login session. Log out and back in, or run\n",{"type":48,"tag":63,"props":1987,"children":1989},{"className":1988},[],[1990],{"type":54,"value":1101},{"type":54,"value":1992}," in each new shell, until the new membership is active.",{"type":48,"tag":57,"props":1994,"children":1995},{},[1996,2001,2003,2008],{"type":48,"tag":658,"props":1997,"children":1998},{},[1999],{"type":54,"value":2000},"Docker runtime still missing",{"type":54,"value":2002},": Restart Docker, then rerun\n",{"type":48,"tag":63,"props":2004,"children":2006},{"className":2005},[],[2007],{"type":54,"value":1027},{"type":54,"value":421},{"type":48,"tag":57,"props":2010,"children":2011},{},[2012,2017,2019,2024,2026,2031,2032,2037,2038,2043,2045,2050],{"type":48,"tag":658,"props":2013,"children":2014},{},[2015],{"type":54,"value":2016},"Driver branch detected != 580",{"type":54,"value":2018},": The driver-branch pin is exact on\ndebian-family (",{"type":48,"tag":63,"props":2020,"children":2022},{"className":2021},[],[2023],{"type":54,"value":1219},{"type":54,"value":2025},"). On rhel-\u002Fsuse-family the script\ninstalls the latest open driver shipped in NVIDIA's CUDA 13.0 repo for\nthe detected distro, which is always ≥ 580. If your host needs a stricter\npin, set ",{"type":48,"tag":63,"props":2027,"children":2029},{"className":2028},[],[2030],{"type":54,"value":1251},{"type":54,"value":136},{"type":48,"tag":63,"props":2033,"children":2035},{"className":2034},[],[2036],{"type":54,"value":1258},{"type":54,"value":159},{"type":48,"tag":63,"props":2039,"children":2041},{"className":2040},[],[2042],{"type":54,"value":1275},{"type":54,"value":2044}," to the exact package names you want before\nrunning ",{"type":48,"tag":63,"props":2046,"children":2048},{"className":2047},[],[2049],{"type":54,"value":238},{"type":54,"value":421},{"type":48,"tag":57,"props":2052,"children":2053},{},[2054,2066,2068,2074],{"type":48,"tag":658,"props":2055,"children":2056},{},[2057,2059,2064],{"type":54,"value":2058},"Driver installed but ",{"type":48,"tag":63,"props":2060,"children":2062},{"className":2061},[],[2063],{"type":54,"value":201},{"type":54,"value":2065}," fails",{"type":54,"value":2067},": Load the module with\n",{"type":48,"tag":63,"props":2069,"children":2071},{"className":2070},[],[2072],{"type":54,"value":2073},"sudo modprobe nvidia",{"type":54,"value":2075}," or reboot. Secure Boot may require MOK enrollment on\nsystems where it is enabled.",{"type":48,"tag":57,"props":2077,"children":2078},{},[2079,2084,2086,2091],{"type":48,"tag":658,"props":2080,"children":2081},{},[2082],{"type":54,"value":2083},"Kubernetes still has no GPU capacity",{"type":54,"value":2085},": Confirm the driver works on each GPU\nnode with ",{"type":48,"tag":63,"props":2087,"children":2089},{"className":2088},[],[2090],{"type":54,"value":201},{"type":54,"value":2092},", then check the GPU Operator\u002Fdevice plugin pods and node\nlabels.",{"type":48,"tag":2094,"props":2095,"children":2096},"style",{},[2097],{"type":54,"value":2098},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":2100,"total":2257},[2101,2119,2137,2148,2160,2174,2187,2201,2214,2223,2237,2246],{"slug":2102,"name":2102,"fn":2103,"description":2104,"org":2105,"tags":2106,"stars":2116,"repoUrl":2117,"updatedAt":2118},"nemoclaw-user-guide","retrieve NemoClaw documentation and configuration","Guides human users' AI agents to the NemoClaw docs MCP server and canonical Fern documentation in Markdown form. Use when users ask how to install, configure, operate, troubleshoot, secure, or learn NemoClaw with an AI coding assistant. Trigger keywords - nemoclaw docs, use nemoclaw with ai agent, nemoclaw mcp docs, nemoclaw install help, nemoclaw quickstart, nemoclaw markdown docs, llms.txt, agent skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2107,2110,2113],{"name":2108,"slug":2109,"type":13},"Documentation","documentation",{"name":2111,"slug":2112,"type":13},"MCP","mcp",{"name":2114,"slug":2115,"type":13},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":2120,"name":2120,"fn":2121,"description":2122,"org":2123,"tags":2124,"stars":2134,"repoUrl":2135,"updatedAt":2136},"mcore-build-and-dependency","manage Megatron-LM development environments","Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2125,2128,2131],{"name":2126,"slug":2127,"type":13},"Containers","containers",{"name":2129,"slug":2130,"type":13},"Deployment","deployment",{"name":2132,"slug":2133,"type":13},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":2138,"name":2138,"fn":2139,"description":2140,"org":2141,"tags":2142,"stars":2134,"repoUrl":2135,"updatedAt":2147},"mcore-bump-base-image","update NVIDIA PyTorch base images","Bump the NVIDIA PyTorch base image (`nvcr.io\u002Fnvidia\u002Fpytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker\u002F.ngc_version.dev` and GitLab CI in `.gitlab\u002Fstages\u002F01.build.yml`), the post-bump CI loop (re-run functional tests, refresh golden values, mark broken tests), and the gotchas that bit PRs",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2143,2146],{"name":2144,"slug":2145,"type":13},"CI\u002FCD","ci-cd",{"name":2129,"slug":2130,"type":13},"2026-07-14T05:25:59.97109",{"slug":2149,"name":2149,"fn":2150,"description":2151,"org":2152,"tags":2153,"stars":2134,"repoUrl":2135,"updatedAt":2159},"mcore-cicd","manage CI\u002FCD pipelines for Megatron-LM","CI\u002FCD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request\u002FBRANCH ref — always dry-run and verify the destination first; never run against shared or protected branches), and CI failure investigation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2154,2155,2156],{"name":2144,"slug":2145,"type":13},{"name":2129,"slug":2130,"type":13},{"name":2157,"slug":2158,"type":13},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":2161,"name":2161,"fn":2162,"description":2163,"org":2164,"tags":2165,"stars":2134,"repoUrl":2135,"updatedAt":2173},"mcore-create-issue","investigate CI failures and create issues","Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2166,2169,2170],{"name":2167,"slug":2168,"type":13},"Debugging","debugging",{"name":2157,"slug":2158,"type":13},{"name":2171,"slug":2172,"type":13},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":2175,"name":2175,"fn":2176,"description":2177,"org":2178,"tags":2179,"stars":2134,"repoUrl":2135,"updatedAt":2186},"mcore-linting-and-formatting","lint and format Megatron-LM code","Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2180,2183],{"name":2181,"slug":2182,"type":13},"Best Practices","best-practices",{"name":2184,"slug":2185,"type":13},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":2188,"name":2188,"fn":2189,"description":2190,"org":2191,"tags":2192,"stars":2134,"repoUrl":2135,"updatedAt":2200},"mcore-migrate-gpt-to-hybrid","migrate Megatron-LM models to HybridModel","Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2193,2196,2199],{"name":2194,"slug":2195,"type":13},"Machine Learning","machine-learning",{"name":2197,"slug":2198,"type":13},"Migration","migration",{"name":9,"slug":8,"type":13},"2026-07-17T06:07:11.777011",{"slug":2202,"name":2202,"fn":2203,"description":2204,"org":2205,"tags":2206,"stars":2134,"repoUrl":2135,"updatedAt":2213},"mcore-onboard-gb200-1node-tests","onboard functional tests for GB200","Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2207,2210],{"name":2208,"slug":2209,"type":13},"QA","qa",{"name":2211,"slug":2212,"type":13},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":2215,"name":2215,"fn":2216,"description":2217,"org":2218,"tags":2219,"stars":2134,"repoUrl":2135,"updatedAt":2222},"mcore-run-on-slurm","launch distributed training jobs on SLURM","How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2220,2221],{"name":2129,"slug":2130,"type":13},{"name":21,"slug":22,"type":13},"2026-07-14T05:25:49.362534",{"slug":2224,"name":2224,"fn":2225,"description":2226,"org":2227,"tags":2228,"stars":2134,"repoUrl":2135,"updatedAt":2236},"mcore-split-pr","split pull requests to reduce review load","Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2229,2232,2233],{"name":2230,"slug":2231,"type":13},"Code Review","code-review",{"name":2157,"slug":2158,"type":13},{"name":2234,"slug":2235,"type":13},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":2238,"name":2238,"fn":2239,"description":2240,"org":2241,"tags":2242,"stars":2134,"repoUrl":2135,"updatedAt":2245},"mcore-testing","run and manage Megatron-LM tests","Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2243,2244],{"name":2208,"slug":2209,"type":13},{"name":2211,"slug":2212,"type":13},"2026-07-14T05:25:54.928983",{"slug":2247,"name":2247,"fn":2248,"description":2249,"org":2250,"tags":2251,"stars":2134,"repoUrl":2135,"updatedAt":2256},"nightly-sync","manage nightly main-to-dev sync workflows","Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2252,2255],{"name":2253,"slug":2254,"type":13},"Automation","automation",{"name":2144,"slug":2145,"type":13},"2026-07-30T05:29:03.275638",496,{"items":2259,"total":2353},[2260,2277,2287,2301,2311,2326,2339],{"slug":2261,"name":2261,"fn":2262,"description":2263,"org":2264,"tags":2265,"stars":23,"repoUrl":24,"updatedAt":2276},"accelerated-computing-cudf","accelerate data processing with cuDF","Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV\u002FParquet I\u002FO, nullable semantics, and multi-GPU DataFrame workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2266,2269,2272,2273],{"name":2267,"slug":2268,"type":13},"Data Analysis","data-analysis",{"name":2270,"slug":2271,"type":13},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":13},{"name":2274,"slug":2275,"type":13},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":2278,"name":2278,"fn":2279,"description":2280,"org":2281,"tags":2282,"stars":23,"repoUrl":24,"updatedAt":2286},"aiq-deploy","deploy and manage NVIDIA AI-Q infrastructure","Use when asked to install, deploy, run, validate, troubleshoot, or stop NVIDIA AI-Q Blueprint infrastructure.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2283,2284,2285],{"name":2129,"slug":2130,"type":13},{"name":21,"slug":22,"type":13},{"name":9,"slug":8,"type":13},"2026-07-14T05:29:06.667109",{"slug":2288,"name":2288,"fn":2289,"description":2290,"org":2291,"tags":2292,"stars":23,"repoUrl":24,"updatedAt":2300},"aiq-research","conduct deep research with AI-Q","Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2293,2296,2297],{"name":2294,"slug":2295,"type":13},"Agents","agents",{"name":9,"slug":8,"type":13},{"name":2298,"slug":2299,"type":13},"Research","research","2026-07-14T05:28:06.816956",{"slug":2302,"name":2302,"fn":2303,"description":2304,"org":2305,"tags":2306,"stars":23,"repoUrl":24,"updatedAt":2310},"amc-run-sample-calibration","run AMC sample dataset calibration","Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2307,2308,2309],{"name":2267,"slug":2268,"type":13},{"name":9,"slug":8,"type":13},{"name":2211,"slug":2212,"type":13},"2026-07-17T05:29:03.913266",{"slug":2312,"name":2312,"fn":2313,"description":2314,"org":2315,"tags":2316,"stars":23,"repoUrl":24,"updatedAt":2325},"amc-run-video-calibration","calibrate video datasets with AutoMagicCalib","Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API. Use when user has local MP4s and says 'calibrate my videos', 'run AMC on these videos', or similar. For RTSP\u002Flive streams, use amc-run-rtsp-calibration instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2317,2318,2321,2322],{"name":2253,"slug":2254,"type":13},{"name":2319,"slug":2320,"type":13},"Imaging","imaging",{"name":9,"slug":8,"type":13},{"name":2323,"slug":2324,"type":13},"Video","video","2026-07-17T05:28:53.905004",{"slug":2327,"name":2327,"fn":2328,"description":2329,"org":2330,"tags":2331,"stars":23,"repoUrl":24,"updatedAt":2338},"amc-setup-calibration-stack","deploy AutoMagicCalib microservice with Docker","Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2332,2333,2334,2335],{"name":2129,"slug":2130,"type":13},{"name":15,"slug":16,"type":13},{"name":9,"slug":8,"type":13},{"name":2336,"slug":2337,"type":13},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":2340,"name":2340,"fn":2341,"description":2342,"org":2343,"tags":2344,"stars":23,"repoUrl":24,"updatedAt":2352},"cudaq-guide","develop quantum applications with CUDA-Q","CUDA-Q onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[2345,2346,2349],{"name":9,"slug":8,"type":13},{"name":2347,"slug":2348,"type":13},"Quantum Computing","quantum-computing",{"name":2350,"slug":2351,"type":13},"Simulation","simulation","2026-07-14T05:26:58.898253",305]