print(torch.backends.cudnn.enabled) -> FALSE

Hello

I am trying to run

but i couldnt get torch working with cuda,(gives “AssertionError: CUDA is not available” error)
and for my best knowledge i did everything right,
but i am not a programmer so that could be a false assumption.
(got Cudnn,Cuda installed)

print(torch.cuda.is_available())

FALSE

print(torch.backends.cudnn.enabled)

True

print(torch.backends.cudnn.version())

8302

print(torch.version.cuda)

11.3

print(torch.cuda.device_count())

0

x = torch.rand(5, 3)
print(x)
tensor([[0.4047, 0.7574, 0.3756],
[0.6907, 0.2448, 0.9255],
[0.8376, 0.6428, 0.0788],
[0.4214, 0.3840, 0.7919],
[0.7861, 0.1685, 0.6711]])

in the command prompt nvcc -V gives me this

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Sun_Mar_21_19:24:09_Pacific_Daylight_Time_2021
Cuda compilation tools, release 11.3, V11.3.58
Build cuda_11.3.r11.3/compiler.29745058_0

running nvidia-smi gives me this

±----------------------------------------------------------------------------+
| NVIDIA-SMI 516.94 Driver Version: 516.94 CUDA Version: 11.7 |
|-------------------------------±---------------------±---------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce … WDDM | 00000000:26:00.0 On | N/A |
| 28% 62C P0 41W / 180W | 778MiB / 8192MiB | 2% Default |
| | | N/A |
±------------------------------±---------------------±---------------------+

Anyone have an idea why torch doesnt recognize i got cuda installed?

I eventually solved it the problem was →
I use Substance Painter for work, and it needs some registry changes
(GPU drivers crash with long computations (TDR crash) | Substance 3D Painter)
i missclicked, and setup the Tldlevel value with a Qword, and not a Dword.
(this one little change disabled my cuda,opencl, and physx support)
also use GPU-z to validate if your cuda is working or not.