RTX 5060 Ti 安装 TensorFlow

RTX 5060 has compute capability 12.0 - a very new architecture. Stable TensorFlow versions lack pre-compiled GPU kernels for this architecture, causing JIT compilation failures for certain operations (especially float32).

Step by Step Solution

conda create --name tf_gpu python=3.11
conda activate tf_gpu

conda install nvidia/label/cuda-12.5.1::cuda-toolkit

pip install tf-nightly[and-cuda]

conda env config vars set LD_LIBRARY_PATH=$CONDA_PREFIX/lib:$LD_LIBRARY_PATH

conda deactivate
conda activate tf_gpu



Enjoy Reading This Article?

Here are some more articles you might like to read next:

  • LaTeX Math Typesetting Guide
  • Shortcuts Cheat Sheet
  • tmux User Guides
  • Ubuntu 22.04 MacOS Monterey 主题
  • CASIA-WebMaskedFace 模拟佩戴口罩人脸数据集