coreos/fedora-bootc-nvidia

coreos 프로필

최근 기록된 소스 활동: 2026년 3월 19일 08:36

SkillsMP 카탈로그 업데이트: 2026년 10월 17일 00:09

수집된 skills: 1

GitHub 스타: 11
GitHub 포크: 4

저장소 README

README 원본 파일: README.md

README 원본 보기

README 전체 펼치기

# Fedora bootc with NVIDIA drivers and tools **STATUS:** **Work In Progress** — This image is in active development and is not yet ready for production use. We are deriving Fedora bootc images to integrate NVIDIA components (kernel modules, drivers, etc). This specialized image is hosted on Quay.io at [quay.io/coreos-devel/fedora-bootc-nvidia](https://quay.io/repository/coreos-devel/fedora-bootc-nvidia?tab=tags). The build process utilizes official NVIDIA DKMS packages from [NVIDIA repo](https://developer.download.nvidia.com/compute/cuda/repos/) to compile the open kernel modules, a method usually reserved for the client system. Despite requiring a tweak to target the immutable image's specific kernel (rather than the build host's), using DKMS simplifies integration significantly. However, this introduces the downside of breaking Secure Boot as the kmods are not signed by the Fedora's key but a self-signed one, which is ok for a POC. For permanent deployment, we'd need to maintain a custom signing key, which adds the extra step of enrolling it with MOK to regain secure booting functionality. ## Building locally ```bash source build-args.conf podman build --build-arg-file build-args.conf -f Containerfile.builder -t $BUILDER_IMAGE podman build --build-arg-file build-args.conf -f Containerfile -t localhost/fedora-bootc-nvidia ``` ## To test it ```bash [core@localhost ~]$ lsmod | grep nvidia nvidia_drm 155648 0 nvidia_modeset 2248704 1 nvidia_drm nvidia 15917056 1 nvidia_modeset drm_ttm_helper 16384 1 nvidia_drm video 81920 1 nvidia_modeset [core@localhost ~]$ lspci -nnk | grep -A2 NVIDIA 00:03.0 VGA compatible controller [0300]: NVIDIA Corporation GA102 [GeForce RTX 3080] [10de:2206] (rev a1) Subsystem: NVIDIA Corporation GA102 [GeForce RTX 3080] [10de:1467] Kernel driver in use: nvidia Kernel modules: nouveau, nvidia_drm, nvidia # Checking the licenses [core@localhost ~]$ modinfo -l nvidia nvidia_drm nvidia_modeset Dual MIT/GPL Dual MIT/GPL Dual MIT/GPL ``` ## Serve a LLM with RamaLama Boot an existing Image Mode system, for example Fedora CoreOS: Create a Containerfile to layer the the NVIDIA CUDA driver and libs: ``` cat > Containerfile << 'EOF' ARG STREAM=44 ARG VERSION=610.43.02 FROM quay.io/coreos-devel/fedora-bootc-nvidia:${STREAM}-${VERSION} RUN <<EORUN set -xeuo pipefail source /usr/lib/os-release curl https://developer.download.nvidia.com/compute/cuda/repos/${ID}${VERSION_ID}/$(arch)/cuda-${ID}${VERSION_ID}.repo \ -o /etc/yum.repos.d/cuda.repo dnf install -y \ nvidia-driver-cuda dnf clean all systemctl enable nvidia-persistenced nvidia-cdi-refresh EORUN RUN bootc container lint EOF ``` Build that container locally ``` sudo podman build -t fedora-bootc-nvidia . ``` Then switch the system to that build ``` sudo bootc switch --transport containers-storage localhost/fedora-bootc-nvidia --apply ``` Check if everything is ok: ``` nvidia-smi ``` And finally, serve the requested LLM with RamaLama ``` ramalama --engine podman serve -p 8081 --oci-runtime crun --image quay.io/ramalama/cuda:0.12.4 mistral:7b-v3 ```

저장소 README를 불러오는 중…

이 저장소의 skills

네트워크·컴퓨터 시스템 관리자

직업 카테고리 1개 · 100% 분류됨

skill

직업 분류

설명

업데이트

update-fedora-bootc-nvidia

네트워크·컴퓨터 시스템 관리자

Guide to update the fedora-bootc-nvidia image to a new Fedora version and/or NVIDIA driver

원문 언어: 영어

2026-03-19

수집된 skill 1개 중 1개를 표시합니다.

AI 어시스턴트로 설치

이 프롬프트를 복사해 사용 중인 AI 어시스턴트에 입력하세요.

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/coreos/fedora-bootc-nvidia.