Tensorflow can't detect GPU in linux aarch64 environment

Hi all,

I have been trying to use the NVIDIA Geforce RTX 3070 in my linux system(Ubuntu 22.04.4 LTS) for Tensorflow model training. Unfortunately, Tensorflow couldn’t detect the GPU. I have tried updating all the drivers and libraries.

List of drivers installed for GPU:

nvidia-driver-535-server, (kernel modules provided by linux-modules-nvidia-535-server-generic)
nvidia-driver-515, (kernel modules provided by nvidia-dkms-515)
nvidia-driver-520, (kernel modules provided by nvidia-dkms-520)
nvidia-driver-555-open, (kernel modules provided by nvidia-dkms-555-open)
nvidia-driver-535, (kernel modules provided by linux-modules-nvidia-535-generic)
nvidia-driver-545, (kernel modules provided by nvidia-dkms-545)
nvidia-driver-525, (kernel modules provided by nvidia-dkms-525)
nvidia-driver-555, (kernel modules provided by nvidia-dkms-555)
nvidia-driver-550, (kernel modules provided by linux-modules-nvidia-550-generic)
nvidia-driver-545-open, (kernel modules provided by nvidia-dkms-545-open)
nvidia-driver-535-server-open, (kernel modules provided by linux-modules-nvidia-535-server-open-generic)
nvidia-driver-550-open, (kernel modules provided by linux-modules-nvidia-550-open-generic)
nvidia-driver-535-open, (kernel modules provided by linux-modules-nvidia-535-open-generic)

Installed python libraries:

tensorflow 2.13.1
tensorflow-aarch64 2.16.1
nvidia-cublas-cu12 12.4.2.65
nvidia-cudnn-cu12 9.1.1.17
nvidia-dali-cuda120 1.37.1
nvidia-dali-tf-plugin-cuda120 1.37.1
nvidia-nvimgcodec-cu12 0.2.0.7
nvidia-pyindex 1.0.9

After all the installtions, I have setup the path also:

CUDNN_PATH=$(dirname $(python3 -c “import nvidia.cudnn;print(nvidia.cudnn.file)”))
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CONDA_PREFIX/lib/:$CUDNN_PATH/lib

But when I try to get the GPU devices in my PC:
python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"

output is []

I don’t know If I’m missing something here. So I hope to find someone to help me with this issue. What else can I do?

Hi @raghul, This might be due to the nvidia drivers incompatibility. Could you please try to downgrade Nvidia drivers 552.44 and also Tensorflow 2.13 supports CUDA 11.8. but I can see that you are using CUDA 12.5 which is incompatible. Thank You.