Environment :
- 3050 Ti
- Cuda 12.5 (latest drivers)
- Windows 11
- Miniconda 3
I’ve been trying to make PyTorch Cuda work for two days but I have been unable to.
torch.cuda.is_available()
Will always return false.
Nvidia-smi returns
I have simply ran this command :
conda install pytorch torchvision torchaudio pytorch-cuda=12.4 -c pytorch -c nvidia
I have tried different versions (12.1, 12.4) and also tried to reinstall my NVidia drivers or the miniconda environment.
I am trying to run this on my laptop which has 2 GPU, one from the NVIDIA card with Cuda and one from the AMD CPU/GPU for low energy consumption.
When I try to get the environment information it is missing Cuda for Pytorch :
Collecting environment information…
PyTorch version: 2.3.0
Is debug build: False
CUDA used to build PyTorch: Could not collect
ROCM used to build PyTorch: N/AOS: Microsoft Windows 11 Pro
GCC version: Could not collect
Clang version: Could not collect
CMake version: Could not collect
Libc version: N/APython version: 3.12.3 | packaged by Anaconda, Inc. | (main, May 6 2024, 19:42:21) [MSC v.1916 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-11-10.0.22631-SP0
Is CUDA available: False
CUDA runtime version: 12.5.40
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: GPU 0: NVIDIA GeForce RTX 3050 Ti Laptop GPU
Nvidia driver version: 555.85
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
I can run software that uses Cuda without issue (cuda-z or a few graphics softwares) so I know the drivers doesn’t have an issue.
The only time I managed somehow to have the environment setup properly, any Cuda calls would hang forever.
Is there a problem of version ? What can I do to make this work ?