Installing Julia on jetson nano is it worth it?

New to Julia
1

I have a jetson nano and want to load up linux and add julia to work on cuda coding. I have read that I should NOT install the Julia binaries but to MAKE the julia environment and then it works perfectly. The issue I have is that the whole process will take 56 hours and that concerns me. I have two questions:

1 how long did it take you to install a julia development environment on a jetson 4gb nano?

2 what were the steps you followed to get that result?

thank you

2

I have not tried Julia on a Jetson Nano, but the official 1.6.2 arm64 binary works fine on a Raspberry Pi 4 4GB. I would try installing the arm64 binary and see what happens, I bet it just works.

The Pi has A72 cores which are the slightly newer but mostly identical upgrade to the A57 in the Jetson.

3

You can see SystemBenchmark results for the Jetson Xavier NX here

https://github.com/IanButterworth/SystemBenchmark.jl/issues/8#issuecomment-626410322

I had some earlier experience with the nano and precompilation time (before parallel precomp existed) was excruciatingly slow

4

thanks @contradict GREAT idea! didn’t think about that. Worth a shot. Thanks so much.

5

HI there @ianshmean
great work on the benchmarks but the NX is out of my price range right now. I want to practice on the Nano first and then move up a notch if I get the hang of it. thanks so much for the benchmarks.

6

But the Jetsons have GPU’s, and the RPI’s do not.
Which is why I want Julia running on the Jetsons so badly.