Windows Installer or Docker?

Bob98

I am trying to optimize processing on my Dell XPS15 laptop. I've got 64GB RAM, two SSD drives and the Nvidia GPU. I am striving for the highest quality output along with the best performance. I have been down the docker / WSL route and while I have had some success I went down the Windows installer path as well. I am finding conflicting and/or old information about GPU support - lots of statements saying it only works in the docker installation but there's the release of the installer indicating is supports the gpu. For new installations on relatively small systems like mine what is the recommended approach? I'd rather spend more time learning WebODM than docker, linux, wsl2 etc.

4 Replies

Bob98 OP

I am working with projects that have 400-700 images. I have been able to get them to process but I suspect I will run into memory issues again as I tune parameters. I would like to think a swap file on the SSD drive should handle memory issues - run slower but should still complete. I also want to expose WebODM on my local network so I can access it from other computers.
I understand Docker Desktop does not support gpu acceleration either. So, I have configured docker to run without the desktop. I've also seen statements that WebODM's gpu is acceleration is Linux based and therefore has to run in a Docker environment.

Bob98 OP

And then I suppose you get into the resources used by Docker itself, passing data through WSL to docker or whatever ... when you consider those overhead issues is the gain from using the NVIDIA gpu worth the effort or is it actually offset by non-native windows operations resulting in slower performance?

poltrojan

I'm curious if wouldn't be better to set it up on cloud sever utilizing server hardware memory and cpu rather than local of your laptops; 400-700 images is a lot to process.

pierotofy

Windows Installer supports GPU 👍

Need more help?

This discussion happened in the WebODM community Discord. Join and ask your own question.

Join View Original Thread

This page mirrors a public discussion from the WebODM Discord. How to request a removal.