Docker uses containers tocreate virtual environments that isolate a TensorFlow installation from the restof the system. TensorFlow programs are run within this virtual environment thatcan share resources with its host machine (access directories, use the GPU,connect to the Internet, etc.). TheTensorFlow Docker images are tested for each release.
Docker is the easiest way to enable TensorFlow GPU support on Linux since only theNVIDIA® GPU driver is required on the host machine (the NVIDIA® CUDA® Toolkit does not need tobe installed).
TensorFlow Docker requirements
- Make sure you have installed the NVIDIA driver and Docker engine for your Linux distribution Note that you do not need to install the CUDA Toolkit on the host system, but the NVIDIA driver needs to be installed. For instructions on getting started with the NVIDIA Container Toolkit, refer to the installation guide.
- Docker uses containers to create virtual environments that isolate a TensorFlow installation from the rest of the system. TensorFlow programs are run within this virtual environment that can share resources with its host machine (access directories, use the GPU, connect to the Internet, etc.).
- @cdpath $ sudo apt install -y nvidia-docker2 Reading package lists. Done Building dependency tree Reading state information. Done E: Unable to locate package nvidia-docker2.
- Nvidia-docker is a wrapper around NVIDIA Container Runtime which registers the NVIDIA runtime by default and provides the nvidia-docker command. To use nvidia-docker, install the nvidia-docker AUR package and then restart docker. Containers with NVIDIA GPU support can then be run using any of the following methods: # docker run -runtime=nvidia.
- Install Docker onyour local host machine.
- For GPU support on Linux, install NVIDIA Docker support.
- Take note of your Docker version with
docker -v. Versions earlier than 19.03 require nvidia-docker2 and the
--runtime=nvidiaflag. On versions including and after 19.03, you will use the
nvidia-container-toolkitpackage and the
--gpus allflag. Both options are documented on the page linked above.
- Take note of your Docker version with
sudo, create the
dockergroup andadd your user. For details, see thepost-installation steps for Linux.
NVIDIA Drivers¶. Before you get started, make sure you have installed the NVIDIA driver for your Linux distribution. The recommended way to install drivers is to use the package manager for your distribution but other installer mechanisms are also available (e.g. By downloading.run installers from NVIDIA Driver Downloads).
Download a TensorFlow Docker image
The official TensorFlow Docker images are located in the tensorflow/tensorflow Docker Hub repository. Image releases are tagged using the following format:
|The latest release of TensorFlow CPU binary image. Default.|
|Nightly builds of the TensorFlow image. (Unstable.)|
|Specify the version of the TensorFlow binary image, for example: 2.1.0|
|Nightly builds of a TensorFlow |
|Special experimental image for developing TF custom ops. More info here.|
Each base tag has variants that add or change functionality:
|The specified tag release with GPU support. (See below)|
|The specified tag release with Jupyter (includes TensorFlow tutorial notebooks)|
You can use multiple variants at once. For example, the following downloadsTensorFlow release images to your machine:
Start a TensorFlow Docker container
To start a TensorFlow-configured container, use the following command form:
For details, see the docker run reference.
Examples using CPU-only images
Let's verify the TensorFlow installation using the
latest tagged image. Dockerdownloads a new TensorFlow image the first time it is run:
Let's demonstrate some more TensorFlow Docker recipes. Start a
bash shellsession within a TensorFlow-configured container:
Within the container, you can start a
python session and import TensorFlow.
To run a TensorFlow program developed on the host machine within a container,mount the host directory and change the container's working directory(
-v hostDir:containerDir -w workDir):
Permission issues can arise when files created within a container are exposed tothe host. It's usually best to edit files on the host system.
Start a Jupyter Notebook server usingTensorFlow's nightly build:
Follow the instructions and open the URL in your host web browser:
Docker is the easiest way to run TensorFlow on a GPU since the host machineonly requires the NVIDIA® driver (the NVIDIA® CUDA® Toolkit is not required).
Install the Nvidia Container Toolkit to add NVIDIA® GPU support to Docker.
nvidia-container-runtime is onlyavailable for Linux. See the
nvidia-container-runtimeplatform support FAQ for details.
Check if a GPU is available:
nvidia-dockerv1 uses the
nvidia-dockeralias, rather than the
--gpus allcommand line flags.
Examples using GPU-enabled images
Ubuntu Docker Install Nvidia Driver
Download and run a GPU-enabled TensorFlow image (may take a few minutes):
It can take a while to set up the GPU-enabled image. If repeatedly runningGPU-based scripts, you can use
docker exec to reuse a container.
Home page for Docker's documentation. Getting Started with Docker. Take a walkthrough that covers writing your first app, data storage, networking, and swarms, and ends with your app running on production servers in the cloud. Docker cli windows download. Windows Server 2016 is the where Docker Windows containers should be deployed for production. For developers planning to do lots of Docker Windows container development, it may also be worth setting up a Windows Server 2016 dev system (in a VM, for example), at least until Windows 10 and Docker for Windows support for Windows containers matures. To install Docker for older versions of Windows, see Install Docker Toolbox on Windows. Configure your shared drives. The AWS SAM CLI requires that the project directory, or any parent directory, is listed in a shared drive. The Docker Desktop installation includes Docker Engine, Docker CLI client, Docker Compose, Notary, Kubernetes, and Credential Helper. Install and run Docker Desktop on Mac Double-click Docker.dmg to open the installer, then drag the Docker icon to the Applications folder.
Use the latest TensorFlow GPU image to start a
bash shell session in the container:
CUDA on WSL User Guide
The guide for using NVIDIA CUDA on Windows Subsystem for Linux.
Windows Subsystem for Linux (WSL) is a Windows 10 feature that enables users to run native Linux command-line tools directly on Windows. WSL is a containerized environment within which users can run Linux native applications from the command line of the Windows 10 shell without requiring the complexity of a dual boot environment. Internally, WSL is tightly integrated with the Microsoft Windows operating system, which allows it to run Linux applications alongside traditional Windows desktop and modern store apps.
With WSL 2 and GPU paravirtualization technology, Microsoft enables developers to run GPU accelerated applications on Windows.The following document describes a workflow for getting started with running CUDA applications or containers in a WSL 2 environment.
Getting started with running CUDA on WSL requires you to complete these steps in order:
Installing Microsoft Windows Insider Program Builds
Install the latest builds from the Microsoft Windows Insider Program
Register for the Microsoft Windows Insider Program.
Install the latest build from the Dev Channel.Note:
Ensure that you install Build version 20145 or higher. We recommend being on WIP OS 21332 and higher with Linux Kernel 5.4.91+ for the best performance.
You can check your build version number by running winver via the Windows Run command.
Installing NVIDIA Drivers
Download the NVIDIA Driver from the download section on the CUDA on WSL page. Choose the appropriate driver depending on the type of NVIDIA GPU in your system - GeForce and Quadro.
Install the driver using the executable. This is the only driver you need to install.
The DirectX WSL driver is installed automatically along with other driver components so no additional action is needed for installation. This driver enables graphics on WSL2.0 by supporting DX12 APIs. TensorFlow with DirectML support on WSL will get NV GPU hardware acceleration for training and inference workloads. There are no present capabilities in WSL, hence the driver is oriented towards compute/machine learning tasks. For some helpful examples, see https://docs.microsoft.com/en-us/windows/win32/direct3d12/gpu-tensorflow-wsl.
Do not install any Linux display driver in WSL. The Windows Display Driver will install both the regular driver components for native Windows and for WSL support.Note:
NVIDIA is aware of a specific installation issue reported on mobile platforms with the WIP driver 465.12 posted on 11/16/2020. A known workaround will be to disable and reenable the GPU adapter from device manager at system start. We are working on a fix for this issue and will have an updated driver soon.
As an alternative, users may opt to roll back to an earlier driver from device manager driver updates.
Installing WSL 2
This section includes details about installing WSL 2, including setting up a Linux distribution of your choice from the Microsoft Store.
- Install WSL 2 by following the instructions in the Microsoft documentation available here.
- Ensure you have the latest kernel by clicking “Check for updates” in the “Windows Update” section of the Settings app. If the right update with the kernel 4.19.121+ is installed, you should be able to see it in the Windows Update history. Alternatively, you can check the version number by running the following command in PowerShell:
- If you don’t see this update, then in the Windows Update Advanced options, make sure to enable recommended Microsoft updates and run the check again:
- If you don’t have the last WSL kernel updated, you will see the following blocking warning upon trying to launch a Linux distribution within WSL 2.
- Launch the Linux distribution and make sure it runs in WSL 2 mode using the following command:
Setting up CUDA Toolkit
It is recommended to use the Linux package manager to install the CUDA for the Linux distributions supported under WSL 2. Follow these instructions to install the CUDA Toolkit.
First, set up the CUDA network repository. The instructions shown here are for Ubuntu 18.04. See the CUDA Linux Installation Guide for more information on other distributions.
Now install CUDA. Note that for WSL 2, you should use the cuda-toolkit-<version> meta-package to avoid installing the NVIDIA driver that is typically bundled with the toolkit. You can also install other components of the toolkit by choosing the right meta-package.
Do not choose the cuda, cuda-11-0, or cuda-drivers meta-packages under WSL 2 since these packages will result in an attempt to install the Linux NVIDIA driver under WSL 2.
Running CUDA Applications
Just run your CUDA app as you would run it under Linux! Once the driver is installed there is nothing more to do to run existing CUDA applications that were built on Linux.
A snippet of running the BlackScholes Linux application from the CUDA samples is shown below.
Build the CUDA samples available under /usr/local/cuda/samples from your installation of the CUDA Toolkit in the previous section. The BlackScholes application is located under /usr/local/cuda/samples/4_Finance/BlackScholes. Alternatively, you can transfer a binary built on Linux to WSL 2!
Setting up to Run Containers
This chapter describes the workflow for setting up the NVIDIA Container Toolkit in preparation for running GPU accelerated containers.
Use the Docker installation script to install Docker for your choice of WSL 2 Linux distribution. Note that NVIDIA Container Toolkit does not yet support Docker Desktop WSL 2 backend.
Install Nvidia Drivers Docker
Install NVIDIA Container Toolkit
Now install the NVIDIA Container Toolkit (previously known as nvidia-docker2). WSL 2 support is available starting with nvidia-docker2 v2.3 and the underlying runtime library (libnvidia-container >= 1.2.0-rc.1).
For brevity, the installation instructions provided here are for Ubuntu 18.04 LTS.
Setup the stable and experimental repositories and the GPG key. The changes to the runtime to support WSL 2 are available in the experimental repository.
Install the NVIDIA runtime packages (and their dependencies) after updating the package listing.
Open a separate WSL 2 window and start the Docker daemon again using the following commands to complete the installation.
Running CUDA Containers
In this section, we will walk through some examples of running GPU containers in a WSL 2 environment.
Simple CUDA Containers
In this example, let’s run an N-body simulation CUDA sample. This example has already been containerized and available from NGC.
From the console, you should see an output as shown below.
In this example, let’s run Jupyter notebook.
After the container starts, you can see the following output on the console.
After the URL is available from the console output, input the URL into your browser to start developing with the Jupyter notebook. Ensure that you replace 127.0.0.1 with localhost in the URL when connecting to the Jupyter notebook from the browser.
If you navigate to the Cell menu and select the Run All item, then check the log within the Jupyter notebook WSL 2 container to see the work accelerated by the GPU of your Windows PC.
Deep Learning Framework Containers
In this example, let’s run a TensorFlow container to do a ResNet-50 training run using GPUs using the 20.03 container from NGC. This is done by launching the container and then running the training script from the nvidia-examples directory.
Let's look at another example from Lesson 15 of the Learning TensorFlow tutorial. In this example, the code creates a random matrix with a given size as input and then does a element wise operation on the input tensor.
The example also allows you to observe the speedup when the code is run on the GPU. The source code is shown below.
Save the code as matmul.py on the host's C drive, which is mapped as /mnt/c in WSL 2. Run the code using the same 20.03 TensorFlow container in the previous example. The results of running this script, launched from the mounted drive C, on a GPU and a CPU are shown below. For simplicity the output is reduced.
The same example is now run on the CPU.
Get started quickly with AI training using pre-trained models available from NVIDIA and the NGC catalog. Follow the instructions in this post for more details.
- 3/23/2021: Resolved issues with nvidia-smi crashing on some systems.
- NVIDIA Driver for Windows 10: 470.14
- WIP build: 21332, WSL Linux kernel 5.4.91
- WIP OS 21313+ and Linux kernel 5.4.91
- Do not use WIP OS 21327
- 1/28/2021: CUDA Toolkit 11.2 support, nvidia-smi, NVML support and critical performance improvements. The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 465.42
- Recommended WIP build: 21292
- 12/16/2020: Support for 3060Ti. Fix for installation problems observed in notebooks. The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 465.21
- 11/16/2020: The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 465.12
- 9/23/2020: The following software versions are supported with this preview release for WSL 2:
- 9/2/2020: The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 460.15
6/19/2020: Updated driver release to address cache coherency issues on some CPU systems, including AMD Ryzen.The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 455.41
6/17/2020: Initial Version.The following software versions are supported with this preview release for WSL 2:
- NVIDIA Driver for Windows 10: 455.38
- NVIDIA Container Toolkit: nvidia-docker2 (2.3) and libnvidia-container (>= 1.2.0-rc.1)
7.1. New Features
The following new features are included in this release:
- CUDA Toolkit 11.2 support
- Critical performance improvements
- nvidia-smi packaged
- Full NVML support
- IPC support
7.2. Resolved IssuesThe following issues are resolved in this release:
- Fixed nvidia-smi issues on some systems.
7.3. Known Limitations
The following features are not supported in this release:
- Note that NVIDIA Container Toolkit does not yet support Docker Desktop WSL 2 backend. Use Docker-CE for Linux instead inside your WSL 2 Linux distribution.
- nvidia-smi is now supported but in order to use it, please copy it to /usr/bin and set appropriate permissions with the below commands: We will soon provide a better user experience when there is OS support.
- CUDA debugging or profiling tools are not supported in WSL 2. This capability will be added in a future release.
- cumemmap IPC with fd is now supported. Other Legacy IPC APIs are not yet supported.
- Unified Memory is limited to the same feature set as on native Windows systems.
- With the NVIDIA Container Toolkit for Docker 19.03, only --gpus all is supported. This means that on multi-GPU systems it is not possible to filter for specific GPU devices by using specific index numbers to enumerate GPUs.
- When installing CUDA using the package manager, do not use the cuda, cuda-11-0, or cuda-drivers meta-packages under WSL 2. These packages have dependencies on the NVIDIA driver and the package manager will attempt to install the NVIDIA Linux driver which may result in issues. The NVIDIA Windows 10 driver should be the only driver present in the system.
- When running the NGC Deep Learning (DL) Framework GPU containers in WSL 2, you may encounter a message: Note that this message is an incorrect warning for WSL 2 and will be fixed in future releases of the DL Framework containers to correctly detect the NVIDIA GPUs. The DL Framework containers will still continue to be accelerated using CUDA on WSL 2.
7.4. Known Issues
The following are known issues in this release:
- CUDA on WSL2 is not to be used with the latest Microsoft Windows 10 Insider Preview Build 20226 due to known issues. Please use newer builds >= 20236 or revert to the older build 20221 to use CUDA on WSL2.
- Warning: Please do not update to WIP OS 21327 if you would like to use WSL with GPU support.
Here is a collection of potential errors that you may encounter when using CUDA on WSL 2:
Container Runtime Initialization Errors
- In some cases, when running a Docker container, you may encounter nvidia-container-cli : initialization error:
- This usually indicates that the right Microsoft Windows Insider Preview Builds, WSL 2, NVIDIA drivers and NVIDIA Container Toolkit may not be installed correctly. Use the dxdiag tools from the Run dialog and provide the diagnostic logs to NVIDIA. You can also use the CUDA on WSL 2 Forums to get in touch with NVIDIA product and engineering teams for help.
This document is provided for information purposes only and shall not be regarded as a warranty of a certain functionality, condition, or quality of a product. NVIDIA Corporation (“NVIDIA”) makes no representations or warranties, expressed or implied, as to the accuracy or completeness of the information contained in this document and assumes no responsibility for any errors contained herein. NVIDIA shall have no liability for the consequences or use of such information or for any infringement of patents or other rights of third parties that may result from its use. This document is not a commitment to develop, release, or deliver any Material (defined below), code, or functionality.
NVIDIA reserves the right to make corrections, modifications, enhancements, improvements, and any other changes to this document, at any time without notice.
Customer should obtain the latest relevant information before placing orders and should verify that such information is current and complete.
NVIDIA products are sold subject to the NVIDIA standard terms and conditions of sale supplied at the time of order acknowledgement, unless otherwise agreed in an individual sales agreement signed by authorized representatives of NVIDIA and customer (“Terms of Sale”). NVIDIA hereby expressly objects to applying any customer general terms and conditions with regards to the purchase of the NVIDIA product referenced in this document. No contractual obligations are formed either directly or indirectly by this document.
NVIDIA products are not designed, authorized, or warranted to be suitable for use in medical, military, aircraft, space, or life support equipment, nor in applications where failure or malfunction of the NVIDIA product can reasonably be expected to result in personal injury, death, or property or environmental damage. NVIDIA accepts no liability for inclusion and/or use of NVIDIA products in such equipment or applications and therefore such inclusion and/or use is at customer’s own risk.
NVIDIA makes no representation or warranty that products based on this document will be suitable for any specified use. Testing of all parameters of each product is not necessarily performed by NVIDIA. It is customer’s sole responsibility to evaluate and determine the applicability of any information contained in this document, ensure the product is suitable and fit for the application planned by customer, and perform the necessary testing for the application in order to avoid a default of the application or the product. Weaknesses in customer’s product designs may affect the quality and reliability of the NVIDIA product and may result in additional or different conditions and/or requirements beyond those contained in this document. NVIDIA accepts no liability related to any default, damage, costs, or problem which may be based on or attributable to: (i) the use of the NVIDIA product in any manner that is contrary to this document or (ii) customer product designs.
No license, either expressed or implied, is granted under any NVIDIA patent right, copyright, or other NVIDIA intellectual property right under this document. Information published by NVIDIA regarding third-party products or services does not constitute a license from NVIDIA to use such products or services or a warranty or endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual property rights of the third party, or a license from NVIDIA under the patents or other intellectual property rights of NVIDIA.
Reproduction of information in this document is permissible only if approved in advance by NVIDIA in writing, reproduced without alteration and in full compliance with all applicable export laws and regulations, and accompanied by all associated conditions, limitations, and notices.
THIS DOCUMENT AND ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND OTHER DOCUMENTS (TOGETHER AND SEPARATELY, “MATERIALS”) ARE BEING PROVIDED “AS IS.” NVIDIA MAKES NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO THE MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT, MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE. TO THE EXTENT NOT PROHIBITED BY LAW, IN NO EVENT WILL NVIDIA BE LIABLE FOR ANY DAMAGES, INCLUDING WITHOUT LIMITATION ANY DIRECT, INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE, OR CONSEQUENTIAL DAMAGES, HOWEVER CAUSED AND REGARDLESS OF THE THEORY OF LIABILITY, ARISING OUT OF ANY USE OF THIS DOCUMENT, EVEN IF NVIDIA HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Notwithstanding any damages that customer might incur for any reason whatsoever, NVIDIA’s aggregate and cumulative liability towards customer for the products described herein shall be limited in accordance with the Terms of Sale for the product.
DisplayPort and DisplayPort Compliance Logo, DisplayPort Compliance Logo for Dual-mode Sources, and DisplayPort Compliance Logo for Active Cables are trademarks owned by the Video Electronics Standards Association in the United States and other countries.
HDMI, the HDMI logo, and High-Definition Multimedia Interface are trademarks or registered trademarks of HDMI Licensing LLC.
OpenCL is a trademark of Apple Inc. used under license to the Khronos Group Inc.
NVIDIA and the NVIDIA logo are trademarks or registered trademarks of NVIDIA Corporation in the U.S. and other countries. Other company and product names may be trademarks of the respective companies with which they are associated.
© 2007-2021 NVIDIA Corporation. All rights reserved.
This product includes software developed by the Syncro Soft SRL (http://www.sync.ro/).