RAM usage is definitely the biggest problem with Julia on a Pi Zero. It's always swapping.Dealbreaker: RAM use? Ridiculous to assume all programs will involve matrix multiplication, and therefore reserve a big chunk of RAM for openBLAS just in case. I can understand a bigger system image with JIT/LLVM involved but the Julia people actually believe that achieving a 200MB startup image is something to celebrate! It wouldn't take many instances to burn through a Pi RAM.
From a different point of view Julia is a domain specific language. Making vectors and matrices a built-in data type allows the language to include convenient Matlab style syntax for working with vectors and matrices.
Statistics: Posted by ejolson — Tue Feb 04, 2025 12:51 am