I was making simulations with Flux.jl, DiffEqFlux.jl, Optim.jl, DiffEqSensitivity.jl and OrdinaryDiffEq.jl in Julia-1.5.2 and I saw that there was a new stable release (1.6.2 version), so I did upgrade Julia to the ‘new’ one (and removed everything about 1.5.2 version), nevertheless I realized the running time (making exactly the same simulations) was around 12x slower than before. Anybody knows why this happens or knows how to recover an old version of Julia (1.5.2) on Mac?
To use a different version on linux (Mac should be quite similar), I:
Go to the downloads page and download the version you want.
tar zxvf julia-latest-linuxaarch64.tar.gz
mv julia-latest-XXX/ julia-2021-XX-XX
Then in .profile (your username’s .profile and root’s profile), add:
Then after you reboot and launch Julia, you will get that version.
If using scripts, you can use the shebang at the top of the script:
#!/usr/bin/env julia (to use the version you specified in .profile)
#!/home/XXX/.julia/julia-1.0.5/bin/julia (to use a specific version only for that script)
Hi again @tortijazz
Just in case you are not familiar with system administration things, I realised I should be more specific, in .profile you still need to keep what is already there (or you will bork your system), so it should be something like:
Also, if you want to run your program with specific versions of packages, you can look at setting up a
Brett-mahar, thank you very much. You helped me a lot.
It would be helpful if you could narrow down what is slower in Julia 1.6.2. If you are adventurous, it will also be interesting to try the nightly builds to see if the issue may have already been resolved.
Gustavo, I’m happy I could help!