Tensorflow cuda version compatibility table. x is compatible with CUDA 12.

Tensorflow cuda version compatibility table CUDA/cuDNN version. 10) are equivalent and they can interoperate (models trained in one work in the other without any concern). Usually, TensorFlow versions are tightly aligned with specific CUDA and cuDNN versions. La compatibilidad con GPU de TensorFlow requiere una selección de controladores y bibliotecas. Does an overview of the compatible versions or even a list of officially tested combinations exist? I can't find it in the TensorFlow documentation. 0 or higher version. 2(TensorFlow 2. Just get tensorflow to run, that's the hard part. 14 the CUDA version supported is 11. 9 Bazel version TensorFlow 1. 04 LTS 64-bit Python Version (if applicable): 3. x requires CUDA 11. I have noticed that some newer TensorFlow versions are incompatible with older CUDA and cuDNN versions. x) is just a wrapper on top of tf. To fix this issue you should follow tested build configurations. The following images and the link provide an overview of the officially supported/tested combinations of CUDA and TensorFlow on Linux, macOS and Windows: Jul 3, 2024 · This document is for users who need backwards compatibility across different versions of TensorFlow (either for code or data), and for developers who want to modify TensorFlow while preserving compatibility. 0 tensorflow-gpu=1. Reference the TensorFlow compatibility table available on TensorFlow's documentation site for guidance Jan 13, 2022 · Is the CUDA version mentioned in the list, the maximum version that can be installed or minimum version that should be installed? Neither. 3은 GraphDef 버전 8을 추가하고 버전 4부터 8까지 지원할 수 있습니다. 0 cuDNN バージョン。 For a complete list of supported drivers, see the CUDA Application Compatibility topic. 1 or not? Dec 2, 2024 · If a serialized engine was created using the version-compatible flag, it could run with newer versions of TensorRT within the same major version. Provide details and share your research! But avoid …. For a complete list of supported drivers, see CUDA Application Compatibility. May 17, 2022 · @hontimzam,. Thus, users should upgrade from all R418, R440, R450, R460, R510, R520, R530, R545 and R555 drivers, which are not forward-compatible with CUDA 12. 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. 105_418. test. 1 and typed in the following code "import tensorflow as tf" Then "tf. Nota: La compatibilidad con GPU está disponible para Ubuntu y Windows con tarjetas habilitadas para CUDA®. 14 build with CUDA 11. 6. Nov 19, 2024 · Check for Proper TensorFlow and CUDA Versions . 3 PyTorch Version (if applicable): NA Baremetal or Container (if container which image + tag Jul 31, 2024 · CUDA 11. 2 for tensorflow 2. 2, python version is 3. Apr 20, 2024 · The following sections highlight the compatibility of NVIDIA ® cuDNN versions with the various supported NVIDIA CUDA ® Toolkit, CUDA driver, and NVIDIA hardware versions. NVIDIA GPUs power millions of desktops, notebooks, workstations and supercomputers around the world, accelerating computationally-intensive tasks for consumers, professionals, scientists, and researchers. Yes. This can be done using pip: pip uninstall tensorflow pip install tensorflow-gpu Dec 11, 2020 · I think 1. Dec 20, 2024 · Uninstall the existing CPU version of TensorFlow, then install the GPU version. Feb 5, 2024 · I have written few codes where I need tensorflow (using gpu) and cudf, cupy all together. You need a CUDA-enabled NVIDIA GPU and Python installed. No response. I have googled this, searched every question but I never got answers for Feb 8, 2020 · In order to use tensorflow-gpu with CUDA 10. *|9. Thank you! We call a SavedModel which was created using only non-deprecated, non-experimental, non-compatibility APIs in TensorFlow major version N a SavedModel supported in version N. Using mismatched versions can lead to runtime errors, performance degradation, and incorrect results. 1. /install_tf. x is compatible with CUDA 12. Dec 9, 2024 · Installation of Tensorflow2 with GPU support is easy and the only complication can be arisen from the CUDA compability which in turns depends on the Nvidia driver version. 1を避けるなど） tensorflowとtensorflow-gpuがダブっていないか？ tensorflow-gpuとpython系は同じバージョンでインストールされているか？ 動かしたいソースコードはどのバージョンで作られているか？ Aug 15, 2018 · Keras - Tensorflow versions compatibility is a frequent problem that i have faced many times myself. Nov 23, 2024 · Nvidia driver and CUDA version compatibility chart - nvidia_driver_cuda_version_compatibility_chart. 1 Feb 16, 2021 · dear @anax32 I am trying to install tensorflow-gpu=1. I remember that i installed my tensorflow using wheel. Jan 7, 2025 · CUDA Version: The TensorRT detector utilizes the 12. Or you can try dockerizing each mutually incompatible part and create a pipeline from that, although I am not sure when you would need to do this instead just separa Aug 28, 2021 · The tensorflow which I was using , is not utilizing gpu now, after it has been changed accidentally by other person. Compute Capability: The GPU must support a Compute Capability of 5. 2 or newer. Dec 17, 2024 · However, this constant evolution can lead to version compatibility issues, especially when migrating older projects to newer TensorFlow versions. 0 이상)를 지원합니다. 39 (Windows) as indicated, minor version compatibility is possible across the CUDA 11. Missing DLLs on Windows: Make sure that CUDA and cuDNN paths are correctly added to your system’s environment variables. *'] " and i have tried to search for a compatible version on anaconda. Jul 21, 2024 · Issue type Build/Install Have you reproduced the bug with TensorFlow Nightly? No Source source TensorFlow version 2. 6 by mistake. 2 or Oct 8, 2019 · 表のとおりにバージョンを合わせたか？（CUDA=9ならば9. TensorFlow releases are tested against specific CUDA versions. 0은 버전 4부터 7까지에 대한 지원을 중단하고 버전 8만 남길 수 있습니다. 11 and later include experimental support for Singularity v3. Installation can be little tricky, but there are lots of how on the internet to install Nvidia driver, Cuda, cuDNN, NCCL and so on Share If you're working with a code based on the old framework version, you must check what driver version you need to install. Deep learning framework containers 19. 0 Custom code Yes OS platform and distribution Ubuntu 22. However, the functionality required to build or modify Are you looking for the compute capability for your GPU, then check the tables below. x releases that ship after this cuDNN release. GPU, CUDA Toolkit, and CUDA Driver Requirements Mar 28, 2021 · I had installed CUDA 10. The compatibility between TensorFlow and CUDA versions is as follows: The CUDA driver's compatibility package only supports particular drivers. js TensorFlow Lite TFX All libraries RESOURCES Models & datasets Tools Responsible AI Recommendation systems Groups Contribute Blog Forum Nov 12, 2023 · Find out your Cuda version by running nvidia-smi in terminal. 13. Jul 5, 2023 · For your GPU to work you need to have the correct versions of CUDA, cuDNN, tensorflow and python. Currently, the version of Cuda is 10. 08 is based on TensorFlow 2. 12 python=3. org but nothing exist i need a way to install them I even tried to use pip Mar 17, 2022 · All versions of Tensorflow (as in, the specific 2. GPU Requirements Release 21. For more information, see CUDA Compatibility and Upgrades and NVIDIA CUDA and Drivers Support. Nov 20, 2024 · This guide is for the latest stable version of TensorFlow. Install TensorFlow. 2 nécessite la version 450. May 24, 2023 · Going forward for the upcoming releases, tensorflow might move to CUDA 12. 15. 1+ (opset version 7 and higher). 0 – pip uninstall tensorflow # remove current version pip install /mnt/tensorflow-version-tags. 0 -> cudatoolkit=8. Mar 10, 2019 · GPU : RTX2070, Tensorflow version : 1. X build for python 3. NVIDIA driver: 545. Use the legacy kernel module flavor. 7 and tensorflow version is 2. Configuraciones de compilación probadas Linux CPU Feb 15, 2022 · Hi @sushreebarsa according the compatibility table, all the tensorflow doesnot support cuda11. gpu_devices_name() and my gpu shows up just fine (it didnt last time). 5 still "supports" cc3. 02 (ou version ultérieure) CUDA® Toolkit : TensorFlow (2. Supported Operator Data Types Nov 8, 2018 · In your case: yes you can use latest version of tensorflow. 9 TensorFlow Version (if applicable): 1. 0 The python command for package using PIP is pip install tensorflow-gpu==1. 0 and later. Oct 29, 2024 · For GPU support, set cuda=Y during configuration and specify the versions of CUDA and cuDNN. 2 tensorflow tensorflow can only use CPU and can not utilize GPU. For older container versions, refer to the Frameworks Support Matrix. Jul 31, 2018 · Check the CUDA version: and cuDNN version: and install a combination as given below in the images or here. The cuDNN build for CUDA 12. 0 using conda but I get "Package cudatoolkit conflicts for: cudnn=7. I think there may be a compatibility issue here. Install the Cuda Toolkit for your Cuda version. 16. The output of the following code gives me:. x. Environment. The CUDA driver's compatibility package only supports particular drivers. 02 以降が必要です。 CUDA® ツールキット - TensorFlow は CUDA® 11. May 1, 2023 · Learn how to solve the common problem of GPU not detected in TensorFlow with CUDA by checking compatibility, verifying installation, enabling device, updating driver, and updating Windows. For example: if an ONNX Runtime release implements ONNX opset 9, it can run models stamped with ONNX opset versions in the range [7-9]. 04 Mobile device No response Python version 3. Windows 10. x family of toolkits. NVIDIA TensorFlow Container Versions. A guide to understanding the compatibility of TensorFlow and CUDA versions, with a recommended table of compatible combinations. 7 Bazel version No response GCC/Compile Loaded runtime CuDNN library: 7102 (compatibility version 7100) but source was compiled with 7005 (compatibility version 7000). 2에는 450. 4 would be the last PyTorch version supporting CUDA9. For more information, see CUDA Compatibility and Upgrades. 0 and this link: cannot train Keras convolution network on GPU but changing cuDNN version doesn't work for me. 0 - CUDA version: 12. 2; Bibliothèque CUPTI fournie avec le CUDA® Toolkit; SDK cuDNN 8. What is the expected version compatibility rules for TensorRT? I didn't have any luck finding any documentation on that. 3. 0 Custom Code No OS Platform and Distribution Linux Ubuntu 16. sh <desired_tensorflow_version>, replacing <desired_tensorflow_version> with the version of the TensorFlow you want to install (without version - download 2. TensorRT Version: 8. 6 -> tensorflow-gpu-base==1. keras, which I do not think is that you want, and this is why it requires specifically TensorFlow 2. 0 was released with an earlier driver version, but by upgrading to Tesla Recommended Drivers 450. 2는 GraphDef 버전 4부터 7까지 지원할 수 있습니다. md Oct 19, 2020 · There is a couple of things if you want to upgrade to a new version of tensorflow-gpu: First, in using tensorflow-gpu for having compatible versions together you have to try to install the tensorflow-gpu using the conda package manager. 1, it seems that the version you are using was compiled with CUDA 10. 1 you will have to build tensorflow-gpu with CUDA 10. TensorFlow provides various ways to ensure that your models remain functional and efficient across different versions with the TensorFlow Compat (compat) module. 1 released on (Feb 2019), version 10. Apr 2, 2021 · As of today, there are a lot of versions available for TensorFlow, CUDA and cuDNN, which might confuse the developers or the beginners to select right compatible combination to make their development environment. 10. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11. 06 OS: Zorin 17 (based on Ubuntu 22. Sep 20, 2021 · Thanks for contributing an answer to Stack Overflow! Please be sure to answer the question. NVIDIA® GPU ドライバ - CUDA® 11. 1 yourself or download a pre-compiled version that was build with CUDA 10. TensorFlow container image version 23. TensorFlow container image version 24. Num GPUs Available: 0 Is CUDA available: False Is cuDNN available: True. : Tensorflow-gpu == 1. Depending on the framework you are using (ex. Ensure that the TensorFlow version installed is compatible with the CUDA version on your system. Table 1. It is the exact minor version of the CUDA runtime you must have available. 14. 적어도 6개월 후, TensorFlow 2. 0, cuDNN : 7. Ask Question Asked 4 years, 9 The CUDA driver's compatibility package only supports particular drivers. so. 5 to 2. Sep 21, 2018 · Request: Version-Compatible Table There are a lot of people using Tensorflow, and I'm sure that those people at least have tried to use tflearn, tensorflow-gpu, cuDNN CUDA, etc. To check the TensorFlow version, use the following command in your Aug 15, 2022 · In order to use TensorFlow with CUDA, you need to check the compatibility chart to make sure you’re using compatible versions. GCC/compiler version. 12 -c rapidsai -c conda-forge -c nvidia cudf=23. TensorFlow supports CUDA version 9. Dec 4, 2024 · This guide explains how to install TensorFlow with GPU support on your computer. Jan 9, 2025 · TensorFlow depends on multiple components and the supported features of those components can affect the TensorFlow ROCm supported feature set. Aug 4, 2024 · This command will install the appropriate version of CUDA for your TensorFlow installation. Below you can see compatibility tables for: Tensorflow PyTorch Oct 8, 2022 · Click to expand! Issue Type Support Source binary Tensorflow Version tensorflow-gpu==1. Aug 3, 2022 · To provide some more background, many users already have a certain CUDA version installed and want to know whether some TensorFlow issue they are getting (usually GPU not detected) is caused by using an incompatible CUDA version or because of a different reason. Dec 18, 2024 · Why Version Compatibility Matters. js TensorFlow Lite TFX LIBRARIES TensorFlow. Install the NVIDIA CUDA Toolkit. 2 に対応しています（TensorFlow は 2. 0: cannot open shared object file: No such file or directory Jul 2, 2020 · The problem is that the latest keras version (2. Jul 24, 2022 · Asking a concrete question about writing code using Tensorflow is "specifically about programming". We will announce that information in the mentioned official doc once the next release happens. TensorFlow's performance and your project's reproducibility depend significantly on maintaining a compatible environment. Asking an abstract support question about compatibility between some undefined binary release of Tensorflow and a specific release of CUDA (or the other way around as the question was originally written, which makes even less sense) isn't. Oct 3, 2022 · Notice. Tensorflow, PyTorch), you'll need to install a proper CUDA, cuDNN, and GCC version. 7 in the next TensorFlow container release. 0 cudnn=7. Below are the minimum requirements: CUDA: TensorFlow 2. Differences in CUDA Versions. 1 is CUDA Toolkit 10. Is there a tensorflow version that is compatible with my cuda and cudnn versions? I also need this working tensorflow version to be >=1. 2 and from 2. 29. Note that you don’t need a local CUDA toolkit, if you install the conda binaries or pip wheels, as they will ship with the CUDA runtime. Before, it was not a problem but now, if I install: conda create --solver=libmamba -n rapids-23. 0 or greater, which typically includes Maxwell-era GPUs or newer. 11 the CUDA version supported is 11. In your example: versions > 11. Same for CUDNN (and that CUDNN version must also be version matched to your CUDA runtime). You can learn more about Compute Capability here. 12 to 2. 8 is compatible with tensorflow 1. 5 devices; the R495 driver in CUDA 11. 8. 0 and cuDNN version is 7. If that doesn't work, you need to install drivers for nVidia graphics card first. 0 Pilotes NVIDIA® GPU : CUDA® 11. Current behavior? I wanted to check what is the cuda and cUDNN version NVIDIA® GPU 드라이버 - CUDA® 11. If your system has multiple versions of CUDA or cuDNN installed, explicitly set the version instead of relying on the default. Python version. CUPTI는 CUDA® Toolkit과 함께 제공됩니다. Refer to these tables for older TensorFlow version requirements. 7 (via pyenv) According to this table… Oct 9, 2024 · However, TensorFlow GPU version has specific hardware requirements: NVIDIA GPU: TensorFlow GPU only supports NVIDIA GPUs that are compatible with CUDA. Jul 23, 2016 · I found it how to install Tensorflow-gpu on a compute capability 2. 7. 0, run Dec 18, 2024 · Dealing with GPU Versions. 1) Versions… TensorFlow. Jan 2, 2010 · Version table for Tensorflow, Scann, Tensorflow Recommenders, CUDA - chrisjune/tensorflow-version-compatibility Aug 16, 2022 · Installing TensorFlow on a system with a CUDA-enabled GPU is a two-step process: 1. Thanks. 6 cudatoolkit=9. For a complete list of supported drivers, see the CUDA Application Compatibility topic. Dec 18, 2024 · The versions of TensorFlow, CUDA, and cuDNN must match. 1 NVIDIA GPU: 3080ti NVIDIA Driver Version: 528. Install cuDNN. 0 -> cudatoolkit[version='8. Refer to the official TensorFlow compatibility table to verify this. You can use following configurations (This worked for me - as of 9/10). x series of CUDA libraries, which necessitates a minimum driver version of >=530 on the host system. ; Execute the script by running . 9 cuda-version=11. 3 since the latest supported version is cuda 11. 0 and higher. Refer to the following table to view the list of supported Visual Studio versions for cuDNN. 9. 02 이상이 필요합니다. 2 I tried cuDNN 7. The CUDA driver's compatibility package only supports specific drivers. The following table shows what versions of Ubuntu, CUDA, TensorFlow, and TensorRT are supported in each of the NVIDIA containers Oct 25, 2023 · Hi @Ankita_Das, As per test build configuration there is no tensorflow version that uses CUDA 11. Check this table for the latest Python, cuDNN, and CUDA version supported by each version of TensorFlow. It would seem that keras 2. GPU Requirements Release 22. 04 supports CUDA compute capability 6. 2. I have seen the list of tested TF version and it's CUDA version compatibility Here, but I am having a doubt whether every TF version can be build from source with any CUDA version or not. So, do not install anything in advanced before reviewing the specific versions. 96. We will stop support for Python 2. 02 (Linux) / 452. To install a GPU-enabled version of TensorFlow, you can specify: pip install tensorflow-gpu==2. Dec 4, 2024 · The following tables highlight the compatibility of cuDNN versions with the various supported OS versions. I would also recommend to follow the general spirit of this tutorial, that is to install the Nvidia driver May 21, 2018 · Currently, I need to install gpu version tensorflow on a machine that existing an older version CUDA and cuDNN, and my question is how can I get it done without re-installing CUDA and cuDNN since other package (see pytorch) need it. Column descriptions: Min CC = minimum compute capability that can be specified to nvcc (for that toolkit version) Jun 28, 2019 · Stack Overflow for Teams Where developers & technologists share private knowledge with coworkers; Advertising & Talent Reach devs & technologists worldwide about your product, service or employer brand Is very important to note that this does not follow the recomendations found at the Tensorflow official website. Aug 15, 2024 · Version compatibility; Introduction Tutorials Guide Learn ML TensorFlow (v2. 0 ou version ultérieure) est compatible avec CUDA® 11. sh. 0. sh and copy the provided script into it. Download the shell script file named install_tf. CUDA is NVIDIA’s parallel computing platform and API model. 17. 5. The Cuda Toolkit version that corresponds with tensorflow 1. Linux # Refer to the following table to view the list of supported Linux versions for cuDNN. I am keeping in my bookmarks this compatibility table, with matches of tensorflow and keras versions. This applies to both the dynamic and static builds of cuDNN. 5, CUDA : 10. OS platform and distribution. 2 - GPU: NVIDA GeForce GTX 1660 Super When I run nvidia-smi it clearly shows my GPU on my Linux. 7 vs the one for 3. (New) PyTorch runs on its own bundled and compiled CUDA stack. For a version compatibility table for GPU TensorFlow on Linux, To install GPU TensorFlow with a non-default CUDA version like 9. The versions in the following table refer to the first TensorFlow version where the ROCm library was introduced as a dependency. NVIDIA Optimized Frameworks such as Kaldi, NVIDIA Optimized Deep Learning Framework (powered by Apache MXNet), NVCaffe, PyTorch, and TensorFlow (which includes DLProf and TF-TRT) offer flexibility with designing and training custom (DNNs for machine learning and AI applications. If you plan to use TensorFlow with GPU support, you may need to install a version that matches your CUDA and cuDNN configuration. Asking for help, clarification, or responding to other answers. 12 is based on TensorFlow 2. 0(cuDNN 버전) (선택사항) TensorRT 6. First reason, it is due to the mismatch of CUDA and cuDNN with Tensorflow version. Generally this problem occurs. If it still doesn't work, it would probably be the easiest to just port the model and weights to newer versions. cuDNN SDK 8. 2, cuDNN 8. 0 CUDNN Version: 8. 11. 8 CUDNN Version: 8. Like for example lets say, I want to have tensorflow-2. 1, Python version : 3. 1 GPU Type: GTX-1060 Nvidia Driver Version: 450. 2. For example:- can TF 1. If a serialized engine was created with hardware compatibility mode enabled, it can run on more than one kind of GPU architecture; the specifics depend on the hardware compatibility level used. This will handle the compatibility of cuDnn and cudatoolkit. The guide covers checking compatibility of your GPU, CUDA, and TensorFlow versions, installing CUDA Toolkit and cuDNN library, and verifying the installation. CUDA® Toolkit - TensorFlow는 CUDA® 11. For the preview build (nightly), use the pip package named tf-nightly. list_physical_devices('GPU')))" Operación exitosa: Se instaló TensorFlow. TensorFlow supports only specific versions of CUDA, and it is essential to use a compatible version to ensure proper functioning. Bazel version. Thank You. 4 Operating System + Version: Ubuntu 18. x for all x, including future CUDA 12. TensorFlow 1. 3. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12. 0 (versions cuDNN) Oct 8, 2024 · TensorFlow version. 03 is based on TensorFlow 2. Here's my specs: - Tensorflow version: 2. 1 NVIDIA GeForce 525M for python ,the trick is simple use a archived version of tensorflow, I used 1. config. 0 on NVIDIA GPUs. All versions of ONNX Runtime support ONNX opsets from ONNX v1. Custom code. . 10 is based on TensorFlow 2. Mar 20, 2020 · Is the latest version of TensorFlow (2. The following table lists the compatible versions of CUDA, cuDNN with TensorFlow. TensorFlow GPU TensorFlow is a powerful tool for machine learning, but can be difficult to get started with. whl cd /tmp # don't import from source directory python-c "import tensorflow as tf; print(\"Num GPUs Available: \", len(tf. 0 以降） CUPTI は CUDA® ツールキットに同梱されています。 cuDNN SDK 8. 1 and CUDNN 7. Apr 2, 2023 · † CUDA 11. * version) Dec 18, 2024 · Why TensorFlow Version Compatibility Matters. ; Make the script executable by running chmod +x install_tf. You can check this on TensorFlow’s official documentation site. 2 には 450. The table below shows the supported versions of CUDA and cuDNN for each TensorFlow release: Dec 18, 2024 · CUDA version mismatch: Ensure that the CUDA version installed on your system matches one of the versions supported by your TensorFlow version. 04) Python: 3. then i need to have CUDA 11. What seems to be the problem? GPU TensorFlow uses CUDA. If you mean "Will I be able to run models trained with older versions of the library", the answer is in TF's release notes and is not related to python. From 2. Look up which versions of python, tensorflow, and cuDNN works for your Cuda version here. 0 - 일부 모델에서 추론 처리량과 지연 시간을 향상 Feb 10, 2024 · Version Compatibility Matrix. Dec 12, 2020 · TensorRT Version: 7. 02 CUDA Version: 11. Feb 10, 2024 · Check this table for the latest Python, cuDNN, and CUDA version supported by each version of TensorFlow. 12. 163 Operating System: Windows 10 Python Version (if applicable): Tensorflow Version (if From what I understand all versions of CUDA are backwards compatible, but after going through the whole process of installing CUDA and setting up a virtual environment with TF this happens when I import tensorflow. 80. Key Features and Enhancements This TensorFlow release includes the following key features and enhancements. Feb 12, 2022 · This table shows you what have already worked for a TensorFlow or a CUDA given version. 4. Software, including libraries such as TensorFlow, gets frequent updates that may include improvements, bug fixes, new features, or incompatible changes. Mobile device. 0) compatible with the latest version of CUDA (10. 5 installer does not. ImportError: libcublas. You should use the highest Python you can for the version of TensorFlow NVIDIA TensorFlow Container Versions The following table shows what versions of Ubuntu, CUDA, TensorFlow, and TensorRT are supported in each of the NVIDIA containers for TensorFlow. 24 CUDA Version: 11. 0 So I just installed the mentioned version of CUDA 11. ONNX Runtime supports all opsets from the latest released version of the ONNX spec. GPU model and memory. 2) ? CUDA/TensorFlow compatibility. Any SavedModel supported in TensorFlow major version N can be loaded and executed with TensorFlow major version N+1. 10 supports CUDA compute capability 6. 1 This article explains how to get complete TensorFlow's build environment details, which includes cuda_version, cudnn_version, cuda_compute_capabilities etc. ppdf toig ldhd djffu dlcj avxb rnu efixr mbeesj csrpw