I've been trying to get this to work forever and still no luck
I have:
- GTX 1050 Ti (on Lenovo Legion laptop)
- the laptop also has an Intel UHD Graphics 630 (i'm not sure if maybe this is interfering?)
- Anaconda
- Visual Studio
- Python 3.9.13
- CUDA 11.2
- cuDNN 8.1
- I added these to the PATH: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\bin C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\libnvvp
- finally I installed tensorflow and created its own environment
and I still can't get it to read my GPU
basically followed https://www.youtube.com/watch?v=hHWkvEcDBO0&t=295s
AND I'm still having no luck.
from tensorflow.python.client import device_lib
print(device_lib.list_local_devices())
yields only information on the CPU
Can anyone please help?
CodePudding user response：
You can upgrade tensorflow to 2.0. It should solve your problem.
CodePudding user response：
Check your tensorflow version and compatability with GPU, update your GPU drivers. CUDA 9/10 would do the job.
follow the official tensorflow link: https://www.tensorflow.org/install/pip#windows-native_1
Do all the steps in the same environment in anaconda.