I’d like to use Julia on EEG and fMRI data. I have downloaded on the Ookami supercomputer, i.e., julia-1.8.5-musl-x86_64.tar.gz . However, trying to run the binary bin/julia gives me
./julia: Exec format error. Binary file not executable.
according to the ookami web page, its an Arm based supercomputer at SUNY stony brook, so the x86_64 binaries will certainly not work.
The generic Arm download might work, but to get best performance you will probably have to build Julia from source. I suspect that the administrators of Ookami have either already installed or are planning to install Julia using one of the supercomputer style package managers, and so you really should contact the Ookami administrators, since building from scratch on an “exotic” system might be tricky
This repository contains some material for the “Julia on A64FX” webinar held for users of the Ookami system @ Stony Brook University on 2022-12-08 (video recording).
@Lester_Ingber Perhaps if you could start at a higher level there will be good advice here.
My advice as someone who builds HPC systems. Clearly you have an ambition to run on the OOkami supercomputer. My advice - concentrate on the Julia aspects first. Get some of your data locally onto an x86 or Mac laptop. Or a desktop. Explore what you want to achieve. Learn about Pluto notebooks in Julia - creating and running Pluto notebooks
The next stage is to talk to the team who manage Oookami. Bring cookies.
Talk to them about their storage layout and how data will be ingested into the system,
And what the job scheduling system is and how you can run Julia on the system,
But please take my advice - if you start by setting a goal of running on arm based Oookami today right away two things coudl happen
a) you will give up in frustration at the complexity.
b) you will end up like me and spend 30 years being more interested in systems than doing useful work
@Lester_Ingber May I suggest the following: copy one of your fMRI datasets in DICOM format locally to your laptop. Install Julia. Write a program to read the data.
When you have problems ask here - everyone is very friendly.