Can you install Rapids 0.16 and TF 2.2 in the same conda environment?

Question

I have tried and always get conflicts.

Answer 1

I ran the following commands and was able to successfully install RAPIDS-0.16 and Tensorflow and Tensorflow-gpu 2.2 :

• First, install rapids in a conda env using one of the commands on https://rapids.ai/start.html . I used : conda create -n rapids-0.16 -c rapidsai -c nvidia -c conda-forge -c defaults rapids=0.16 python=3.7 cudatoolkit=10.2

• Activate the conda environment : conda activate rapids-0.16

• Then install tensorflow using pip install tensorflow==2.2

Please check if you are using the right CUDA library version and check if the right Python version is used to install both RAPIDS and Tensorflow.

Answer 2

As of Jan 2023 and considering things change a lot meanwhile, I would suggest to use the default rapids installation https://rapids.ai/start.html#get-rapids containing tensorflow as an option which currently restricts cudatoolkit=11.2:

conda create -n rapids-22.12 -c rapidsai -c conda-forge -c nvidia  \
    rapids=22.12 python=3.9 cudatoolkit=11.2 \
    tensorflow