Cellfinder installation with CUDA 10ΒΆ

If you have CUDA 10 (necessary for the new Turing architecture then pip may not find a compatible version of tensorflow 1.12.2.

This will be the case if you have one of the following GPUs:

  • Titan RTX

  • GeForce RTX 2080 Ti

  • GeForce RTX 2080

  • GeForce RTX 2070

  • GeForce RTX 2060

  • GeForce GTX 1660 Ti

  • GeForce GTX 1660

  • GeForce GTX 1650

  • Quadro RTX 8000

  • Quadro RTX 6000

  • Quadro RTX 5000

  • Quadro RTX 4000

  • Tesla T4

Alternatively, you may just have CUDA 10 installed with an older GPU. This can be checked using nvcc --version. If you see something like: Cuda compilation tools, release 10.0, V10.0.130 then you have CUDA 10 installed.

To install tensorflow with CUDA 10, there is a prebuild wheel for tensorflow 1.12.0 for for python 3.6 here.

Install via:

pip install --no-cache-dir https://github.com/evdcush/TensorFlow-wheels/releases/download/tf-1.12.0-gpu-10.0/tensorflow-1.12.0-cp36-cp36m-linux_x86_64.whl