NVIDIA GeForce RTX 5090 with CUDA capability sm_120 is not compatible with the current PyTorch installation

windows
1

I keep getting this error:

torch\cuda_init_.py:230: UserWarning:
NVIDIA GeForce RTX 5090 with CUDA capability sm_120 is not compatible with the current PyTorch installation.
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.
If you want to use the NVIDIA GeForce RTX 5090 GPU with PyTorch, please check the instructions at Start Locally | PyTorch

Is there a workaround for Windows? I believe there is one for Linux, but as far as I can tell, nothing for Windows users. Just curious if this is being looked into.

2

Yes, you can use WSL in the meantime while we are building the Windows binaries.

3

Thank you for the confirmation. Is there an estimate on when it will be released?

4

Hello, any news about the windows cuda 2.8 version? i have an 5080 and i cant run it :scream: :cold_sweat: Thanks!

5

All nightly PyTorch binaries with CUDA 12.8 and Blackwell support are available now for Linux x86, SBSA, and Windows.

2 Likes
6

Thanks ptrblck! but i keep recieving the warning that dont work :sob:

UserWarning:
NVIDIA GeForce RTX 5080 with CUDA capability sm_120 is not compatible with the current PyTorch installation.
The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.
If you want to use the NVIDIA GeForce RTX 5080 GPU with PyTorch, please check the instructions at Start Locally | PyTorch

7

Any advice its VERY wellcome!!! THANKS!!!

8

Install the nightly binaries by selecting Preview (Nightly) in the install matrix.

2 Likes
9

Captura de pantalla 2025-03-12 233807
Captura de pantalla 2025-03-12 2338071740×955 69.8 KB

Thanks for the help Ptrblck but it dosnt work, look the screenshot :cold_sweat:

10

You need pip3 install -U --pre torch ...

-U (or --upgrade) upgrades the packages you already installed.

(Maybe it’s better to add this on the Start Locally page?)

1 Like
11

The logs tell you that other (older) versions are already found, so uninstall these.

1 Like
12

OMG THANKS!!! you are the best :heart:

1 Like
13

hi
for cuda 12.8
import torch
print(torch.version)
2.7.0
import torchaudio
2.6.0
how i can fix torchaudio to
2.7.0
Or there is no update

14

There is no torchaudio 2.7 yet. See the index URL you use to install torch: https://download.pytorch.org/whl/nightly/torchaudio/