loading...

re: My failed attempt to set up an old MacbookPro as a deep learning workstation VIEW POST

TOP OF THREAD FULL DISCUSSION
re: Good try! - I fiddled with my 2015 MBP in a similar vein, but have landed on Colab and Kaggle, etc, for now, as well. Let's hope we can gain enough...
 

Thanks! I haven't tried Kaggle before, how do you (not) like it in comparison to Colab? Also regarding expansive GPUs, I recently learned about Jetson boards nvidia.com/en-us/autonomous-machin... they are still expensive, but I'm tempted to buy those as they can become full computers. This one developer.nvidia.com/embedded/jets... looks esp interesting to me.

 

One issue to watch out for with Xavier as a workstation is your software's compatibility with ARM. This isn't always a big issue, but sometimes popular packages can be harder to find pre built binaries for when you switch to ARM. I'm thinking VsCode, Godot, in my own recent experience. It might not be an issue for you, just depends on how much of your development tooling you need to run, versus the pure ML portion of your build (NVIDIA seems to have done a good job including pre built libs needed for various ML apps)

Huh, thanks for the headup. I don't think Xavier NX is out yet, but will ping when it comes out & give it a try ;)

 

Wow, the Xavier board is beefy for such a little guy! That's really neat. I'm actually using a Jetson Nano to power my web based Go(Baduk) system. (github.com/Terkwood/BUGOUT/issues/67) I'm not a very good player so it beats me despite its small GPU 😁

Kaggle is nice because it gives 9 hours of high end GPU quota to each user for free, per week. it looks like another nice supplement to Colab for folks like me who are still learning ML. There are a lot of tutorials available from various competition winners and other learners in the community!

Hope to see you post about your experiences with Xavier if you go ahead with a build. Stay distant! 😷 🤜 🦠

Code of Conduct Report abuse