|
如何选择Tensorflow版本?快来看看经过官方测试验证好的Tensorflow与Python、CUDA、cuDNN的版本对应表🧑博主简介:现任阿里巴巴嵌入式技术专家,15年工作经验,深耕嵌入式+人工智能领域,精通嵌入式领域开发、技术管理、简历招聘面试。CSDN优质创作者,提供产品测评、学习辅导、简历面试辅导、毕设辅导、项目开发、C/C++/Java/Python/Linux/AI等方面的服务,如有需要请站内私信或者联系任意文章底部的的VX名片(ID:gylzbk)💬博主粉丝群介绍:①群内初中生、高中生、本科生、研究生、博士生遍布,可互相学习,交流困惑。②热榜top10的常客也在群里,也有数不清的万粉大佬,可以交流写作技巧,上榜经验,涨粉秘籍。③群内也有职场精英,大厂大佬,可交流技术、面试、找工作的经验。④进群免费赠送写作秘籍一份,助你由写作小白晋升为创作大佬。⑤进群赠送CSDN评论防封脚本,送真活跃粉丝,助你提升文章热度。有兴趣的加文末联系方式,备注自己的CSDN昵称,拉你进群,互相学习共同进步。如何选择Tensorflow版本?快来看看经过官方测试验证好的Tensorflow与Python、CUDA、cuDNN的版本对应表🗒️安装说明🗒️使用pip安装TensorFlow📄TensorFlow2软件包现已推出📄旧版TensorFlow📄系统要求📄硬件要求📄相关pip安装包地址Linuxx86LinuxArm64(仅支持CPU)📄macOSx86(仅支持CPU)📄macOSArm64(仅支持CPU)📄Windows系统(仅支持CPU)🗒️Tensorflow与Python、CUDA、cuDNN的版本对应表📄1.Windows系统📃1.1CPU版本📃1.2GPU版本📄2.Linux/Ubuntu系统📃2.1CPU版本📃2.2GPU版本📄3.macOS系统📃3.1CPU版本📃3.2GPU版本🗒️安装说明除了通过源码构建方式安装使用Tensorflow之外,Tensorflow官方还提供了针对Windows、Linux/Ubuntu和macOS系统上,经过充分测试的预构建TensorFlow包。所以,我们可以直接使用pip来安装,省去很多构建过程中的麻烦。🗒️使用pip安装TensorFlow📄TensorFlow2软件包现已推出tensorflow:支持CPU和GPU的最新稳定版(适用于Ubuntu和Windows)tf-nightly:预览build(不稳定)。Ubuntu和Windows均包含GPU支持。📄旧版TensorFlow对于TensorFlow1.x,CPU和GPU软件包是分开的:tensorflow==1.15:仅支持CPU的版本tensorflow-gpu==1.15:支持GPU的版本(适用于Ubuntu和Windows)📄系统要求Python3.6–3.9若要支持Python3.9,需要使用TensorFlow2.5或更高版本。若要支持Python3.8,需要使用TensorFlow2.2或更高版本。pip19.0或更高版本(需要manylinux2010支持)Ubuntu16.04或更高版本(64位)macOS10.12.6(Sierra)或更高版本(64位)(不支持GPU)macOS要求使用pip20.3或更高版本Windows7或更高版本(64位)适用于VisualStudio2015、2017和2019的MicrosoftVisualC++可再发行软件包GPU支持需要使用支持CUDA®的卡(适用于Ubuntu和Windows)⚠️注意:必须使用最新版本的pip,才能安装TensorFlow2。📄硬件要求从TensorFlow1.6开始,二进制文件使用AVX指令,这些指令可能无法在旧版CPU上运行。阅读GPU支持指南,以在Ubuntu或Windows上设置支持CUDA®的GPU卡。📄相关pip安装包地址部分安装方式需要您提供TensorFlowPython软件包的网址。您需要根据Python版本指定网址。Linuxx86版本网址Python3.9(支持GPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.9(仅支持CPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow_cpu-2.16.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.10(支持GPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.10(仅支持CPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow_cpu-2.16.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.11(支持GPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.11(仅支持CPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow_cpu-2.16.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.12(支持GPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whlPython3.12(仅支持CPU)https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow_cpu-2.16.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whlLinuxArm64(仅支持CPU)版本网址Python3.9https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whlPython3.10https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whlPython3.11https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whlPython3.12https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl📄macOSx86(仅支持CPU)版本网址Python3.9https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp39-cp39-macosx_10_15_x86_64.whlPython3.10https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp310-cp310-macosx_10_15_x86_64.whlPython3.11https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp311-cp311-macosx_10_15_x86_64.whlPython3.12https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp312-cp312-macosx_10_15_x86_64.whl📄macOSArm64(仅支持CPU)版本网址Python3.9https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp39-cp39-macosx_12_0_arm64.whlPython3.10https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp310-cp310-macosx_12_0_arm64.whlPython3.11https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp311-cp311-macosx_12_0_arm64.whlPython3.12https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp312-cp312-macosx_12_0_arm64.whl📄Windows系统(仅支持CPU)版本网址Python3.9https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp39-cp39-win_amd64.whlPython3.10https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp310-cp310-win_amd64.whlPython3.11https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp311-cp311-win_amd64.whlPython3.12https://storage.googleapis.com/tensorflow/versions/2.16.1/tensorflow-2.16.1-cp312-cp312-win_amd64.whl上述表格参考的官方页面请点击此处进行访问。由于中文版页面更新不够及时,查看前注意通过右上角按钮切换语言为English。🗒️Tensorflow与Python、CUDA、cuDNN的版本对应表如下版本对应关系摘录自官方网站,都是经过官方测试的构建配置,特整理如下。大家可按需选用。📄1.Windows系统📃1.1CPU版本版本Python版本编译器构建工具tensorflow-2.16.13.9-3.12CLANG17.0.6Bazel6.5.0tensorflow-2.15.03.9-3.112019年MSVCBazel6.1.0tensorflow-2.14.03.9-3.112019年MSVCBazel6.1.0tensorflow-2.12.03.8-3.112019年MSVCBazel5.3.0tensorflow-2.11.03.7-3.102019年MSVCBazel5.3.0tensorflow-2.10.03.7-3.102019年MSVCBazel5.1.1tensorflow-2.9.03.7-3.102019年MSVCBazel5.0.0tensorflow-2.8.03.7-3.102019年MSVCBazel4.2.1tensorflow-2.7.03.7-3.92019年MSVCBazel3.7.2tensorflow-2.6.03.6-3.92019年MSVCBazel3.7.2tensorflow-2.5.03.6-3.92019年MSVCBazel3.7.2tensorflow-2.4.03.6-3.82019年MSVCBazel3.1.0tensorflow-2.3.03.5-3.82019年MSVCBazel3.1.0tensorflow-2.2.03.5-3.82019年MSVCBazel2.0.0tensorflow-2.1.03.5-3.72019年MSVCBazel0.27.1-0.29.1tensorflow-2.0.03.5-3.7微软VC2017Bazel0.26.1tensorflow-1.15.03.5-3.7微软VC2017Bazel0.26.1tensorflow-1.14.03.5-3.7微软VC2017Bazel0.24.1-0.25.2tensorflow-1.13.03.5-3.7MSVC2015更新3Bazel0.19.0-0.21.0tensorflow-1.12.03.5-3.6MSVC2015更新3Bazel0.15.0tensorflow-1.11.03.5-3.6MSVC2015更新3Bazel0.15.0tensorflow-1.10.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.9.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.8.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.7.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.6.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.5.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.4.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.3.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.2.03.5-3.6MSVC2015更新3Cmakev3.6.3tensorflow-1.1.03.5MSVC2015更新3Cmakev3.6.3tensorflow-1.0.03.5MSVC2015更新3Cmakev3.6.3📃1.2GPU版本⚠️注意:原生Windows上的GPU支持仅适用于2.10或更早版本,从TF2.11开始,Windows不支持CUDA构建。要在Windows上使用TensorFlowGPU,您需要在WSL2中构建/安装TensorFlow或将tensorflow-cpu与TensorFlow-DirectML-Plugin一起使用版本Python版本编译器构建工具cuDNN通用计算架构tensorflow_gpu-2.10.03.7-3.102019年MSVCBazel5.1.18.111.2tensorflow_gpu-2.9.03.7-3.102019年MSVCBazel5.0.08.111.2tensorflow_gpu-2.8.03.7-3.102019年MSVCBazel4.2.18.111.2tensorflow_gpu-2.7.03.7-3.92019年MSVCBazel3.7.28.111.2tensorflow_gpu-2.6.03.6-3.92019年MSVCBazel3.7.28.111.2tensorflow_gpu-2.5.03.6-3.92019年MSVCBazel3.7.28.111.2tensorflow_gpu-2.4.03.6-3.82019年MSVCBazel3.1.08.011.0tensorflow_gpu-2.3.03.5-3.82019年MSVCBazel3.1.07.610.1tensorflow_gpu-2.2.03.5-3.82019年MSVCBazel2.0.07.610.1tensorflow_gpu-2.1.03.5-3.72019年MSVCBazel0.27.1-0.29.17.610.1tensorflow_gpu-2.0.03.5-3.7微软VC2017Bazel0.26.17.410tensorflow_gpu-1.15.03.5-3.7微软VC2017Bazel0.26.17.410tensorflow_gpu-1.14.03.5-3.7微软VC2017Bazel0.24.1-0.25.27.410tensorflow_gpu-1.13.03.5-3.7MSVC2015更新3Bazel0.19.0-0.21.07.410tensorflow_gpu-1.12.03.5-3.6MSVC2015更新3Bazel0.15.07.29.0tensorflow_gpu-1.11.03.5-3.6MSVC2015更新3Bazel0.15.079tensorflow_gpu-1.10.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.9.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.8.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.7.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.6.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.5.03.5-3.6MSVC2015更新3Cmakev3.6.379tensorflow_gpu-1.4.03.5-3.6MSVC2015更新3Cmakev3.6.368tensorflow_gpu-1.3.03.5-3.6MSVC2015更新3Cmakev3.6.368tensorflow_gpu-1.2.03.5-3.6MSVC2015更新3Cmakev3.6.35.18tensorflow_gpu-1.1.03.5MSVC2015更新3Cmakev3.6.35.18tensorflow_gpu-1.0.03.5MSVC2015更新3Cmakev3.6.35.18上述表格参考的官方页面请点击此处进行访问。由于中文版页面更新不够及时,查看前注意通过右上角按钮切换语言为English。📄2.Linux/Ubuntu系统⚠️注意:理论上来讲,下述版本对应表是适用于其他Linux系统的,但放在仅在Ubuntu系统上经过了测试。所以大家在安装使用过程中,如果遇到问题,可以考虑切换到Ubuntu系统再尝试一下,可能就没问题了。📃2.1CPU版本版本Python版本编译器构建工具tensorflow-2.16.13.9-3.12Clang17.0.6Bazel6.5.0tensorflow-2.15.03.9-3.11Clang16.0.0Bazel6.1.0tensorflow-2.14.03.9-3.11Clang16.0.0Bazel6.1.0tensorflow-2.13.03.8-3.11Clang16.0.0Bazel5.3.0tensorflow-2.12.03.8-3.11GCC9.3.1Bazel5.3.0tensorflow-2.11.03.7-3.10GCC9.3.1Bazel5.3.0tensorflow-2.10.03.7-3.10GCC9.3.1Bazel5.1.1tensorflow-2.9.03.7-3.10GCC9.3.1Bazel5.0.0tensorflow-2.8.03.7-3.10GCC7.3.1Bazel4.2.1tensorflow-2.7.03.7-3.9GCC7.3.1Bazel3.7.2tensorflow-2.6.03.6-3.9GCC7.3.1Bazel3.7.2tensorflow-2.5.03.6-3.9GCC7.3.1Bazel3.7.2tensorflow-2.4.03.6-3.8GCC7.3.1Bazel3.1.0tensorflow-2.3.03.5-3.8GCC7.3.1Bazel3.1.0tensorflow-2.2.03.5-3.8GCC7.3.1Bazel2.0.0tensorflow-2.1.02.7,3.5-3.7GCC7.3.1Bazel0.27.1tensorflow-2.0.02.7,3.3-3.7GCC7.3.1Bazel0.26.1tensorflow-1.15.02.7,3.3-3.7GCC7.3.1Bazel0.26.1tensorflow-1.14.02.7,3.3-3.7GCC4.8Bazel0.24.1tensorflow-1.13.12.7,3.3-3.7GCC4.8Bazel0.19.2tensorflow-1.12.02.7,3.3-3.6GCC4.8Bazel0.15.0tensorflow-1.11.02.7,3.3-3.6GCC4.8Bazel0.15.0tensorflow-1.10.02.7,3.3-3.6GCC4.8Bazel0.15.0tensorflow-1.9.02.7,3.3-3.6GCC4.8Bazel0.11.0tensorflow-1.8.02.7,3.3-3.6GCC4.8Bazel0.10.0tensorflow-1.7.02.7,3.3-3.6GCC4.8Bazel0.10.0tensorflow-1.6.02.7,3.3-3.6GCC4.8Bazel0.9.0tensorflow-1.5.02.7,3.3-3.6GCC4.8Bazel0.8.0tensorflow-1.4.02.7,3.3-3.6GCC4.8Bazel0.5.4tensorflow-1.3.02.7,3.3-3.6GCC4.8Bazel0.4.5tensorflow-1.2.02.7,3.3-3.6GCC4.8Bazel0.4.5tensorflow-1.1.02.7,3.3-3.6GCC4.8Bazel0.4.2tensorflow-1.0.02.7,3.3-3.6GCC4.8Bazel0.4.2📃2.2GPU版本版本Python版本编译器构建工具cuDNN通用计算架构tensorflow-2.16.13.9-3.12Clang17.0.6Bazel6.5.08.912.3tensorflow-2.15.03.9-3.11Clang16.0.0Bazel6.1.08.912.2tensorflow-2.14.03.9-3.11Clang16.0.0Bazel6.1.08.711.8tensorflow-2.13.03.8-3.11Clang16.0.0Bazel5.3.08.611.8tensorflow-2.12.03.8-3.11GCC9.3.1Bazel5.3.08.611.8tensorflow-2.11.03.7-3.10GCC9.3.1Bazel5.3.08.111.2tensorflow-2.10.03.7-3.10GCC9.3.1Bazel5.1.18.111.2tensorflow-2.9.03.7-3.10GCC9.3.1Bazel5.0.08.111.2tensorflow-2.8.03.7-3.10GCC7.3.1Bazel4.2.18.111.2tensorflow-2.7.03.7-3.9GCC7.3.1Bazel3.7.28.111.2tensorflow-2.6.03.6-3.9GCC7.3.1Bazel3.7.28.111.2tensorflow-2.5.03.6-3.9GCC7.3.1Bazel3.7.28.111.2tensorflow-2.4.03.6-3.8GCC7.3.1Bazel3.1.08.011.0tensorflow-2.3.03.5-3.8GCC7.3.1Bazel3.1.07.610.1tensorflow-2.2.03.5-3.8GCC7.3.1Bazel2.0.07.610.1tensorflow-2.1.02.7,3.5-3.7GCC7.3.1Bazel0.27.17.610.1tensorflow-2.0.02.7,3.3-3.7GCC7.3.1Bazel0.26.17.410.0tensorflow_gpu-1.15.02.7,3.3-3.7GCC7.3.1Bazel0.26.17.410.0tensorflow_gpu-1.14.02.7,3.3-3.7GCC4.8Bazel0.24.17.410.0tensorflow_gpu-1.13.12.7,3.3-3.7GCC4.8Bazel0.19.27.410.0tensorflow_gpu-1.12.02.7,3.3-3.6GCC4.8Bazel0.15.079tensorflow_gpu-1.11.02.7,3.3-3.6GCC4.8Bazel0.15.079tensorflow_gpu-1.10.02.7,3.3-3.6GCC4.8Bazel0.15.079tensorflow_gpu-1.9.02.7,3.3-3.6GCC4.8Bazel0.11.079tensorflow_gpu-1.8.02.7,3.3-3.6GCC4.8Bazel0.10.079tensorflow_gpu-1.7.02.7,3.3-3.6GCC4.8Bazel0.9.079tensorflow_gpu-1.6.02.7,3.3-3.6GCC4.8Bazel0.9.079tensorflow_gpu-1.5.02.7,3.3-3.6GCC4.8Bazel0.8.079tensorflow_gpu-1.4.02.7,3.3-3.6GCC4.8Bazel0.5.468tensorflow_gpu-1.3.02.7,3.3-3.6GCC4.8Bazel0.4.568tensorflow_gpu-1.2.02.7,3.3-3.6GCC4.8Bazel0.4.55.18tensorflow_gpu-1.1.02.7,3.3-3.6GCC4.8Bazel0.4.25.18tensorflow_gpu-1.0.02.7,3.3-3.6GCC4.8Bazel0.4.25.18上述表格参考的官方页面请点击此处进行访问。由于中文版页面更新不够及时,查看前注意通过右上角按钮切换语言为English。📄3.macOS系统📃3.1CPU版本版本Python版本编译器构建工具tensorflow-2.16.13.9-3.12来自xcode13.6的ClangBazel6.5.0tensorflow-2.15.03.9-3.11来自xcode10.15的ClangBazel6.1.0tensorflow-2.14.03.9-3.11来自xcode10.15的ClangBazel6.1.0tensorflow-2.13.03.8-3.11来自xcode10.15的ClangBazel5.3.0tensorflow-2.12.03.8-3.11来自xcode10.15的ClangBazel5.3.0tensorflow-2.11.03.7-3.10来自xcode10.14的ClangBazel5.3.0tensorflow-2.10.03.7-3.10来自xcode10.14的ClangBazel5.1.1tensorflow-2.9.03.7-3.10来自xcode10.14的ClangBazel5.0.0tensorflow-2.8.03.7-3.10来自xcode10.14的ClangBazel4.2.1tensorflow-2.7.03.7-3.9来自xcode10.11的ClangBazel3.7.2tensorflow-2.6.03.6-3.9来自xcode10.11的ClangBazel3.7.2tensorflow-2.5.03.6-3.9来自xcode10.11的ClangBazel3.7.2tensorflow-2.4.03.6-3.8来自xcode10.3的ClangBazel3.1.0tensorflow-2.3.03.5-3.8来自xcode10.1的ClangBazel3.1.0tensorflow-2.2.03.5-3.8来自xcode10.1的ClangBazel2.0.0tensorflow-2.1.02.7,3.5-3.7来自xcode10.1的ClangBazel0.27.1tensorflow-2.0.02.7,3.5-3.7来自xcode10.1的ClangBazel0.27.1tensorflow-2.0.02.7,3.3-3.7来自xcode10.1的ClangBazel0.26.1tensorflow-1.15.02.7,3.3-3.7来自xcode10.1的ClangBazel0.26.1tensorflow-1.14.02.7,3.3-3.7来自xcode的ClangBazel0.24.1tensorflow-1.13.12.7,3.3-3.7来自xcode的ClangBazel0.19.2tensorflow-1.12.02.7,3.3-3.6来自xcode的ClangBazel0.15.0tensorflow-1.11.02.7,3.3-3.6来自xcode的ClangBazel0.15.0tensorflow-1.10.02.7,3.3-3.6来自xcode的ClangBazel0.15.0tensorflow-1.9.02.7,3.3-3.6来自xcode的ClangBazel0.11.0tensorflow-1.8.02.7,3.3-3.6来自xcode的ClangBazel0.10.1tensorflow-1.7.02.7,3.3-3.6来自xcode的ClangBazel0.10.1tensorflow-1.6.02.7,3.3-3.6来自xcode的ClangBazel0.8.1tensorflow-1.5.02.7,3.3-3.6来自xcode的ClangBazel0.8.1tensorflow-1.4.02.7,3.3-3.6来自xcode的ClangBazel0.5.4tensorflow-1.3.02.7,3.3-3.6来自xcode的ClangBazel0.4.5tensorflow-1.2.02.7,3.3-3.6来自xcode的ClangBazel0.4.5tensorflow-1.1.02.7,3.3-3.6来自xcode的ClangBazel0.4.2tensorflow-1.0.02.7,3.3-3.6来自xcode的ClangBazel0.4.2📃3.2GPU版本版本Python版本编译器构建工具cuDNN通用计算架构tensorflow_gpu-1.1.02.7,3.3-3.6来自xcode的ClangBazel0.4.25.18tensorflow_gpu-1.0.02.7,3.3-3.6来自xcode的ClangBazel0.4.25.18上述表格参考的官方页面请点击此处进行访问。由于中文版页面更新不够及时,查看前注意通过右上角按钮切换语言为English。
|
|