Connecting Rented GPU to Blender Addon Step-By-Step Tutorial?

[TheWanderer45]

I managed to get an instance on Vast.ai running, but I could not figure out how to connect it to the Blender addon. I don't have much experience renting a GPU for anything, so I don't really know what I'm doing wrong. I know I'm supposed to open localhost in a browser tab, but sometimes it's a blank page called "Welcome to Docker," and other times, it refuses to connect. I also don't know how to fill out the info in the blender addon. What type do I use? I know I'm supposed to connect using SSH, but there's four different options that involve that. How do I know which one to pick? Furthermore, how do I fill out the info in the blender addon? There are several options for the SSH type for example. How do I find the host, port, user, and remote path? Hovering over these spots in blender doesn't reveal any information on what these are or how to correctly fill them out. Some of my earlier attempts timed out after a console message saying I didn't have an SSH config file yet (and therefore tried using defaults), but after I set one up, the console showed that it didn't find the SSH key or user even though I defined both in the config file and blender addon. How do I know what a correct SSH config file looks like so I know I did it correctly? At this point, I think a step-by-step tutorial would be best as I have no idea what I've been doing wrong. I would prefer an explanation on how to do it with vast.ai because I already spent $10 on credits and wasted about $1 worth of them on failed attempts. I use macOS and blender 5.2.

[ryichando]

We have a step-by-step tutorial video here

https://st-tech.github.io/ppf-contact-solver/blender_addon/tutorial.html

See "Setting Up vast.ai" in the page.

If you want to save your credits I recommend renting a cheap instance like RTX 3070 so you can take your time.

I will update README with this. Please get back to me again if this still does not help.

[ryichando]

Also please setup your SSH key pair and register it via vast.ai console.

This is a very common setup. You should be able to do it if you ask ChatGPT.

How to do that is beyond our documentation coverage.

[alencooper0]

If you’re trying to connect Blender to a rented GPU, I’d get the remote machine set up properly first before messing too much with the addon. Make sure the NVIDIA drivers are installed and that Blender can actually see the GPU. If you’re using a cloud provider like NeevCloud, I’d check the instance details to see what GPU, drivers, CUDA/OptiX support, and remote access options are available. Once that’s working, install the same or a compatible version of Blender on the remote machine and then set up the addon. The exact steps will depend on whether the addon uses SSH, an API, or some kind of remote rendering setup, I’d definitely do a small test render first. No point uploading a huge project only to find out the GPU isn’t being used. You can also watch the GPU usage while rendering to make sure the workload is actually going to the rented GPU. Another thing that can trip you up is missing files. If your project has textures, HDRIs, caches, simulations, or linked assets, make sure those are available on the remote machine too. And yeah, keeping the Blender and addon versions compatible is important. Once the small test works, then you can move on to the full project. It makes troubleshooting a lot easier because you can figure out whether the problem is the GPU, Blender, the addon, or the file setup.