Skip to content

Julia

This comprehensive guide will help you set up and use Julia, a high-level, high-performance programming language.

Adding the Julia Kernel to OnDemand

Open OnDemand provides a graphical interface to the cluster, which provides an interactive user-interface for code development and testing. The OnDemand interface for Borah is available at ondemand.boisestate.edu

In order to use Julia in a Jupyter Notebook through OnDemand, you'll first need to install the IJulia kernel from the command line.

First, load any of the available Julia modules. You can see the available modules using module avail -i julia and load the module using module load julia/<version number here>. Next, load the Julia module, and open a Julia terminal:

module load julia
julia

After these commands, you'll see your prompt change to the Julia terminal:

julia>

Then in the Julia terminal, install the IJulia kernel:

using Pkg
Pkg.add("IJulia")

After this, the Julia kernel is installed for you. (You'll only need to add the IJulia package once.)

Finally, to use Julia in a notebook, navigate to the Jupyter Notebook App on ondemand.boisestate.edu.

Navigate to the Jupyter Notebook App

Once your Jupyter session starts, select the Julia kernel:

Select the Julia kernel

Julia Batch Script Example

Once you've developed and refined your Julia code, you may want to submit it as a batch submission to the scheduler. Following is an example of how to do just that.

  1. Create your script. The example below is called hello_world.jl:

    hello_world.jl
    print("Hello World")

  2. Create your submission script. The example below is called julia-slurm.sh:

    julia-slurm.sh
    #!/bin/bash
#SBATCH -J julia            # job name
#SBATCH -o log_slurm.o%j    # output and error file name (%j expands to jobID)
#SBATCH -n 1                # total number of tasks requested
#SBATCH -N 1                # number of nodes you want to run on
#SBATCH --cpus-per-task 1
#SBATCH -p bsudfq           # queue (partition)
#SBATCH -t 12:00:00         # run time (hh:mm:ss)


# Load the Julia module
module load julia

# Run the example script
julia hello_world.jl

  3. Submit to the scheduler 

    sbatch julia-slurm.sh

Resources