NVIDIA GeForce RTX 5090

nafe_alhmdan · April 10, 2025, 4:04pm

C:\Users\USER\Desktop\RVC1006Nvidia>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
C:\Users\USER\Desktop\RVC1006Nvidia\runtime\lib\site-packages\torch\cuda_init_.py:173: 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_37 sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90 compute_37.
If you want to use the NVIDIA GeForce RTX 5090 GPU with PyTorch, please check the instructions at Start Locally | PyTorch

warnings.warn(incompatible_device_warn.format(device_name, capability, " ".join(arch_list), device_name))
2025-04-10 18:55:22 | INFO | configs.config | Found GPU NVIDIA GeForce RTX 5090
is_half:True, device:cuda:0

ptrblck · April 10, 2025, 4:06pm

Install the latest nightly binaries with CUDA 12.8 and it will work as the Blackwell support requires CUDA >= 12.8.

nafe_alhmdan · April 10, 2025, 4:42pm

Python 3.12.8 (tags/v3.12.8:2dc476b, Dec 3 2024, 19:30:04) [MSC v.1942 64 bit (AMD64)] on win32
Type “help”, “copyright”, “credits” or “license” for more information.

import torch
print(torch.version)
2.8.0.dev20250408+cu128

nafe_alhmdan · April 10, 2025, 4:45pm

It doesn’t work
Same error

ptrblck · April 10, 2025, 4:49pm

The nightly binary itself works as confirmed e.g. here so I don’t know what the issue in your setup is.
The user in the linked post also claimed to have installed the right binary and later claimed:

Upgrading to the latest nightly (with sm_120 support) resolved that

so I also don’t fully understand what happened before.

nafe_alhmdan · April 10, 2025, 5:08pm

C:\Users\USER\Desktop\RVC1006Nvidia>runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897
C:\Users\USER\Desktop\RVC1006Nvidia\runtime\lib\site-packages\torch\cuda_init_.py:173: 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_37 sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90 compute_37.
If you want to use the NVIDIA GeForce RTX 5090 GPU with PyTorch, please check the instructions at Start Locally | PyTorch

C:\Users\USER\Desktop\RVC1006Nvidia>pause
Press any key to continue . . .
The problem could be with the software or the 5090 card.
Because when I change the card to a 4090, it works.
Is there a solution? I’ve tried everything.

ptrblck · April 10, 2025, 5:13pm

You are still using older PyTorch builds so make sure a single binary built with CUDA 12.8 is installed.

OkYes_Anap · June 11, 2025, 11:10am

Same issue
UserWarning:
NVIDIA GeForce RTX 5090 Laptop GPU 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 Laptop GPU GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/

pip3 list
Package Version

einops 0.8.1
filelock 3.18.0
flash_attn 2.7.4.post1
fsspec 2025.5.1
Jinja2 3.1.6
MarkupSafe 3.0.2
mpmath 1.3.0
networkx 3.5
numpy 2.1.2
pillow 11.0.0
pip 25.0.1
setuptools 80.9.0
sympy 1.14.0
torch 2.7.1+cu128
torchaudio 2.7.1+cu128
torchvision 0.22.1+cu128
typing_extensions 4.14.0

python --version
Python 3.12.10

ptrblck · June 11, 2025, 12:27pm

Dame as before: make sure a single and the correct PyTorch binary is used: NVIDIA GeForce RTX 5090 - #7 by ptrblck

OkYes_Anap · June 11, 2025, 1:30pm

How to do this? I already showed you that I have Python and Torch installed. Well, okay, I will attach a screenshot.

ptrblck · June 11, 2025, 1:31pm

Yes, pip list shows the right version but does not account for other binaries found in e.g. the base env.