Linux / ppc64le
Linux / amd64
CUDA is a parallel computing platform and programming model developed by NVIDIA for general computing on graphical processing units (GPUs). With CUDA, developers can dramatically speed up computing applications by harnessing the power of GPUs.
The CUDA Toolkit from NVIDIA provides everything you need to develop GPU-accelerated applications. The CUDA Toolkit includes GPU-accelerated libraries, a compiler, development tools and the CUDA runtime.
The CUDA container images provide an easy-to-use distribution for CUDA supported platforms and architectures.
The images are governed by the following NVIDIA End User License Agreements. By pulling and using the CUDA images, you accept the terms and conditions of these licenses. Since the images may include components licensed under open-source licenses such as GPL, the sources for these components are archived here.
To view the NVIDIA Deep Learning Container license, click here
For more information on CUDA, including the release notes, programming model, APIs and developer tools, visit the CUDA documentation site.
The following tags will no longer be updated.
This may present itself as the following errors.
Reading package lists... Done W: GPG error: http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY A4B469963BF863CC W: The repository 'http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 InRelease' is not signed. N: Data from such a repository can't be authenticated and is therefore potentially dangerous to use. N: See apt-secure(8) manpage for repository creation and user configuration details.
warning: /var/cache/dnf/cuda-fedora32-x86_64-d60aafcddb176bf5/packages/libnvjpeg-11-1-22.214.171.124-1.x86_64.rpm: Header V4 RSA/SHA512 Signature, key ID d42d0685: NOKEY cuda-fedora32-x86_64 23 kB/s | 1.6 kB 00:00 Importing GPG key 0x7FA2AF80: Userid : "cudatools <firstname.lastname@example.org>" Fingerprint: AE09 FE4B BD22 3A84 B2CC FCE3 F60F 4B3D 7FA2 AF80 From : https://developer.download.nvidia.com/compute/cuda/repos/fedora32/x86_64/7fa2af80.pub Is this ok [y/N]: y Key imported successfully Import of key(s) didn't help, wrong key(s)? Public key for libnvjpeg-11-1-126.96.36.199-1.x86_64.rpm is not installed. Failing package is: libnvjpeg-11-1-188.8.131.52-1.x86_64 GPG Keys are configured as: https://developer.download.nvidia.com/compute/cuda/repos/fedora32/x86_64/7fa2af80.pub The downloaded packages were saved in cache until the next successful transaction. You can remove cached packages by executing 'dnf clean packages'. Error: GPG check FAILED
Updated images will be pushed out over the next few days containing the new repo key. Please follow progress using the links below:
It is now possible to build CUDA container images for all supported architectures using Docker Buildkit in one step. See the example script below.
The deprecated image names
nvidia/cuda-ppc64le will remain available, but no longer supported.
The following product pages still exist but will no longer be supported:
The following gitlab repositories will be archived:
The "latest" tag for CUDA, CUDAGL, and OPENGL images has been deprecated on NGC and Docker Hub.
With the removal of the latest tag, the following use case will result in the "manifest unknown" error:
$ docker pull nvidia/cuda Error response from daemon: manifest for nvidia/cuda:latest not found: manifest unknown: manifest unknown
This is not a bug.
Three flavors of images are provided:
base: Includes the CUDA runtime (cudart)
runtime: Builds on the
baseand includes the CUDA math libraries, and NCCL. A
runtimeimage that also includes cuDNN is available.
devel: Builds on the
runtimeand includes headers, development tools for building CUDA images. These images are particularly useful for multi-stage builds.
The Dockerfiles for the images are open-source and licensed under 3-clause BSD. For more information see the Supported Tags section below.
The NVIDIA Container Toolkit for Docker is required to run CUDA images.
For CUDA 10.0,
nvidia-docker2 (v2.1.0) or greater is recommended. It is also recommended to use Docker 19.03.
Read NVIDIA Container Toolkit Frequently Asked Questions to see if the problem has been encountered before.
After it has been determined the problem is not with the NVIDIA runtime, report an issue at the CUDA Container Image Issue Tracker.
Supported tags are updated to the latest CUDA and cuDNN versions. These tags are also periodically updated to fix CVE vulnerabilities.
For a full list of supported tags, click here.
Visit OpenSource @ Nvidia for the GPL sources of the packages contained in the CUDA base image layers.
WARNING: POSSIBLE MISSING IMAGE TAGS
The Cuda image tags for centos7 and 8 may be missing on NGC and Docker Hub. Centos upstream images often fail security scans required by Nvidia before publishing images. Please check https://gitlab-master.nvidia.com/cuda-installer/cuda/-/issues for any security notices!
A list of tags that are no longer supported can be found here