holoscan-install-source

por nvidia

Compilar o Holoscan SDK a partir do código-fonte usando o script ./run incluído. Use apenas quando os pacotes publicados não atenderem às necessidades do usuário.

npx skills add https://github.com/nvidia/skills --skill holoscan-install-source

Holoscan SDK — Build from Source

Purpose

Build the Holoscan SDK from the nvidia-holoscan/holoscan-sdk source tree using its ./run script (which builds inside a Docker container), producing a local install tree consumable as a CMake dependency.

Prerequisites

  • Linux host with NVIDIA GPU + driver (nvidia-smi).
  • git, Docker with NVIDIA Container Toolkit (docker run --gpus all works), and docker-buildx-plugin.
  • ~20 GB free disk for the build container + build/install trees.
  • 10–30 min for a clean first build.

Limitations

  • Only recommended when published packages (Conda / container / apt / wheel) don't fit — debug symbols, custom CMake options, or unsupported configs.
  • Still requires Docker — the ./run script builds inside a container; this is not a true bare-metal build.
  • Cross-compiling to aarch64 needs qemu-user-static on the host.

Step 0: Consult the Official Install Instructions

Always fetch the "Build from Source" section of https://docs.nvidia.com/holoscan/sdk-user-guide/sdk_installation.html (and the linked GitHub README.md / DEVELOP.md for the chosen tag) before building. Extract: required ./run flags for the target architecture and CUDA major, supported branches/tags, any Dockerfile patches called out for the release, and the test names recommended for verification. If the doc disagrees with anything below, the doc wins.

Step 1: Prerequisites

Check that git and Docker (with GPU passthrough) are available:

git --version
docker --version
docker run --rm --gpus all ubuntu:22.04 nvidia-smi
  • If Docker is missing → help install from https://docs.docker.com/engine/install/
  • If GPU passthrough fails → install NVIDIA Container Toolkit:
    curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
    curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list \
      | sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' \
      | sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
    sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit
    sudo nvidia-ctk runtime configure --runtime=docker && sudo systemctl restart docker
    
  • If Docker buildx is missing: sudo apt-get install docker-buildx-plugin

Step 2: Clone the Repository

Clone repo to ~/holoscan/holoscan-sdk if needed

mkdir -p ~/holoscan/
git clone https://github.com/nvidia-holoscan/holoscan-sdk.git
cd ~/holoscan/holoscan-sdk

To build a specific release tag (recommended for stability):

git tag | grep -E '^v[0-9]' | sort -V | tail -5   # list recent tags
git checkout v<VERSION>                             # e.g. v4.1.0

Step 3: Build

The ./run build script handles container creation, CMake configuration, compilation, and install in one step. Warn the user this takes 10–30 minutes on first run (downloads base image + compiles).

./run build

Common options:

FlagPurpose
--type debugDebug build (symbols, no optimization)
--type RelWithDebInfoRelease + debug symbols
--arch aarch64Cross-compile for ARM64 (needs sudo apt install qemu-user-static)
--gpu igpuiGPU build for Jetson/IGX
--dryrunPreview commands without executing

If CMake cache errors occur after changing options:

./run clear_cache && ./run build

Output lands in these folders, and can be retrieved with ./run get_build_dir and ./run get_install_dir

  • Build dir: build-cu<N>-<arch>/
  • Install dir: install-cu<N>-<arch>/.

Step 4: Run Tests

Run the following tests

  • EXAMPLE_CPP_HELLO_WORLD_TEST
  • EXAMPLE_PYTHON_HELLO_WORLD_TEST
  • EXAMPLE_CPP_TENSOR_INTEROP_TEST
  • EXAMPLE_PYTHON_TENSOR_INTEROP_TEST
  • EXAMPLE_CPP_VIDEO_REPLAYER_TEST
  • EXAMPLE_PYTHON_VIDEO_REPLAYER_TEST
./run test

To run all six required tests at once, use a single-quoted regex (the | must be quoted to prevent bash from treating it as a pipe):

./run test --options "-R 'EXAMPLE_CPP_HELLO_WORLD_TEST|EXAMPLE_PYTHON_HELLO_WORLD_TEST|EXAMPLE_CPP_TENSOR_INTEROP_TEST|EXAMPLE_PYTHON_TENSOR_INTEROP_TEST|EXAMPLE_CPP_VIDEO_REPLAYER_TEST|EXAMPLE_PYTHON_VIDEO_REPLAYER_TEST' --output-on-failure"

Run a specific test by name or regex:

./run test --name <test_name>
./run test --options "-R '<regex>' --output-on-failure"
./run test --verbose

Important: Always single-quote the regex string when it contains | — without quotes, bash interprets | as a pipe and the command fails with command not found.

Expected: all tests pass. Note any failures and report them to the user before continuing.

Step 5: Point Applications at the Install Tree

Once built, applications can use the install tree as a CMake dependency. Give the user this path:

/path/to/holoscan-sdk/install-cu<N>-<arch>/

They can set Holoscan_ROOT or CMAKE_PREFIX_PATH to this directory when building their own applications.

Troubleshooting

SymptomFix
bash: <TEST_NAME>: command not found when running testsThe regex contains | — wrap it in single quotes: --options "-R '<regex>'"
CMake cache errors after option change./run clear_cache && ./run build
Docker buildx not foundsudo apt-get install docker-buildx-plugin
GPU not visible inside build containerVerify NVIDIA Container Toolkit and re-run sudo nvidia-ctk runtime configure --runtime=docker
Cross-compile fails (aarch64)Install qemu: sudo apt-get install qemu-user-static

Mais skills de nvidia

compileiq-debug
nvidia
Use quando algo está errado: Search() trava, todas as avaliações retornam INVALID_SCORE, as pontuações não estão melhorando, toda configuração retorna o mesmo número, erros de ptxas…
create-github-pr
nvidia
Crie pull requests do GitHub usando a CLI gh. Use quando o usuário quiser criar um novo PR, enviar código para revisão ou abrir um pull request. Palavras-chave de acionamento -…
nemoclaw-maintainer-cross-issue-sweep
nvidia
Escaneia outras issues abertas para encontrar aquelas que um determinado PR pode também corrigir ou quebrar acidentalmente. Gera oportunidades de correção adjacentes e riscos de contradição com arquivo:linha…
fhir-basics
nvidia
Ensina aos agentes como funcionam as APIs FHIR R4, quais recursos estão disponíveis, como consultá-los com parâmetros de busca e como analisar corretamente todos os formatos de resposta…
compileiq-validate-result
nvidia
Use APÓS a conclusão de uma Pesquisa e ANTES de reivindicar qualquer aceleração ou enviar um ACF. Carrega o CSV dump_results, extrai os K melhores candidatos (objetivo único)…
changelog-audit
nvidia
Auditar o CHANGELOG.md do Warp antes de um lançamento: recuperar entradas perdidas, ordenar por impacto ao usuário, refinar a linguagem das entradas, ajustar quebras de linha e (no modo de branch de lançamento) incrementar comparação…
maintain-dynamic-plugins
nvidia
Manter carregadores de plugins dinâmicos do NeMo Relay, manifestos, SDKs nativos em Rust, protocolo de worker gRPC, SDK de worker Python, documentação, testes e cobertura do fluxo de lançamento
dgx-diagnose
nvidia
Diagnostique problemas comuns do DGX Station GB300 — falhas de CUDA, direcionamento incorreto de GPU, bugs de contêiner vLLM/SGLang, problemas de estado MIG, erros de NVLink/Fabric Manager,…