#why does julia take so much memory?
1 messages · Page 1 of 1 (latest)
You can't reduce the footprint. Julia always loads 3 things into memory when it starts up: (1) the OpenBLAS linear algebra library, (2) the LLVM compiler backend for fast compilation and (3) the documentation for the things it is aware of at startup. Julia is a language for numerical computing, which is why the developers went with this approach. You should not expect to use it in low-performance environments.
Tell us what you're trying to do and maybe we'll be able to help more
i guess that's the tradeoff they have to make for a fast dynamic language.
Isn't it just the julia REPL that takes all the memory? Try checking it when just running julia <your_script>
what possible device would there be that doesn't have 80mb of ram but you still need to use julia interactively
if you're super short on RAM baremodule may also help for your script but I'm not positive about that and that definitely won't work in the REPL
Oh thats very sensible decision.
I was trying to do game of life with 80x80 matrix. An unoptimised one at that
Note v1.10 will lazy load blas. See my post on that PR.
Also, you can already do binary builds that strip out things like metadata and the repl. That was part of the v1.9 release
oh wow thats great to hear
thats great
fast compilation
haha.. yes, but no 😄
it compiles into fast code, but LLVM itself is awfully slow, and LLVM is mostly the reason why it takes 80 MBs of RAM ^^
Thanks everyone here for help, my problem has been cleared
LLVM is mostly the reason why it takes 80 MBs of RAM ^^
Yeah LLVM is like 80, BLAS is like 30, and then you get to like 120 because of what's left. The good thing is that it's clear what the main contributors are.
May i politely ask you how did you learn julia so well?
Oh...so you are the director of modelling and simulation. thats awesome
I will be there one day