TensorFlow 2.14.0 Fails to Detect GPU

[juanmaarg6]

Hi @z-mahmud22. I'm trying to run it on Google Colab, but it doesn't detect the GPU. The Python version is 3.11.11, TensorFlow is 2.14.0, and CUDA is 11.8 (I followed what you explained in #20).

First, with the new requirements.txt, the tensorflow[and-cuda]==2.14.0 part gives me the following error:

Collecting nvidia-cuda-runtime-cu11==11.8.89 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cuda_runtime_cu11-11.8.89-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cublas-cu11==11.11.3.6 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cublas_cu11-11.11.3.6-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cufft-cu11==10.9.0.58 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cufft_cu11-10.9.0.58-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cudnn-cu11==8.7.0.84 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cudnn_cu11-8.7.0.84-py3-none-manylinux1_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-curand-cu11==10.3.0.86 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_curand_cu11-10.3.0.86-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cusolver-cu11==11.4.1.48 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cusolver_cu11-11.4.1.48-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cusparse-cu11==11.7.5.86 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cusparse_cu11-11.7.5.86-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-nccl-cu11==2.16.5 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_nccl_cu11-2.16.5-py3-none-manylinux1_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-cuda-cupti-cu11==11.8.87 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cuda_cupti_cu11-11.8.87-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cuda-nvcc-cu11==11.8.89 (from tensorflow[and-cuda]==2.14.0->-r requirements.txt (line 16))
Using cached nvidia_cuda_nvcc_cu11-11.8.89-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
INFO: pip is looking at multiple versions of tensorflow[and-cuda] to determine which version is compatible with other requirements. This could take a while.
ERROR: Ignored the following yanked versions: 0.0.1
ERROR: Could not find a version that satisfies the requirement tensorrt==8.5.3.1; extra == "and-cuda" (from tensorflow[and-cuda]) (from versions: 0.0.1.dev5, 8.6.1, 8.6.1.post1, 9.0.0.post11.dev1, 9.0.0.post12.dev1, 9.0.1.post11.dev4, 9.0.1.post12.dev4, 9.1.0.post11.dev4, 9.1.0.post12.dev4, 9.2.0.post11.dev5, 9.2.0.post12.dev5, 9.3.0.post11.dev1, 9.3.0.post12.dev1, 10.0.0b6, 10.0.1, 10.1.0, 10.2.0, 10.2.0.post1, 10.3.0, 10.4.0, 10.5.0, 10.6.0, 10.6.0.post1, 10.7.0)
ERROR: No matching distribution found for tensorrt==8.5.3.1; extra == "and-cuda"

I tried modifying the requirements.txt file by setting:

tensorflow==2.14.0
tensorrt==10.5.0

Instead of tensorflow[and-cuda]==2.14.0. With this change, the requirements.txt file doesn't throw any errors, but it still doesn't detect any GPU in Google Colab.

How can I fix this? I urgently need it for a university project. Thank you very much.