2025/02/06/5080-install

[utterances-bot]

Running PyTorch and Triton on the RTX 5080 | Luke’s Blog

I was beyond stoked at the opportunity of getting my hands on a new RTX 5080 to speed up my machine learning developments! Unfortunately, as soon as I connected the new GPU to my workstation I quickly realised that things did not run quite as smoothly as I hoped for. Drivers needed to be re-installed, and machine learning libraries had to be re-downloaded and even built from source. I thought I’d write up a quick how-to for people to potentially save them some hours and to just get going - I’m pretty sure in a few weeks most of these issues won’t be issues anymore with new software releases.

https://webstorms.github.io/2025/02/06/5080-install.html

[jt-zhang]

Thank you for sharing. I have a question and would appreciate it if someone could help with the following: After the last step of running python setup.py install, does Triton support the features of Blackwell for the RTX 5080?

[webstorms]

Glad to be of help! Unfortunately I don't know the answer to this - I'd recommend checking out the Triton GitHub page or their Discord.

[malleshk-web]

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
this is the error when i try to inference on rtx 5080
and my versions are,
python 3.10
torch 7.0+cu118
torchaudio 2.7.0+cu118
cuda version 12.9

and i tried with
torch 2.6.0+cu124
torchaudio 2.6.0+cu124
torchvision 0.21.0+cu124

Not working

[malleshk-web]

torch 2.8.0.dev20250519+cu128
torchaudio 2.6.0.dev20250520+cu128
torchvision 0.22.0.dev20250520+cu128

I installed this version problem solved and working good RTX 5080 for inference

thank you

[Quetzalcohuatl]

as of may 2025, if you have cuda 12.9 you can install the torch 12.8 version and it seems to be working as malleshk said. Just run this command:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128

import torch
x = torch.tensor(1)
x = x.to('cuda')