JuliaTutorial

JuliaTutorial

Prerequisites

Download the Julia executable from your the Julia official website(https://julialang.org/downloads/).

We will use jupyter notebooks for some of the demonstrations. So make sure to download anaconda.

Setting up IJulia

Launch julia either from command line or by launching the executable, and perform following commands.

julia> using Pkg

julia> Pkg.add("IJulia")

julia> Pkg.build("IJulia")

Instantiate the package

Now, we need to download other necessary dependencies. Julia makes this process pretty easy. Navigate to CMTJulia directory, and launch julia from there.

julia> using Pkg

julia> Pkg.activate("./")

julia> Pkg.instantiate("./")

This should start the installation process. Once you are done, you are ready to start exploring the notebooks!

Parallelism

In order to explore thread based parallelism, we need to launch julia in multi-thread mode (https://docs.julialang.org/en/v1.0/manual/parallel-computing/#Multi-Threading-(Experimental)-1). If you are using Linux or MacOS, launch the terminal and do

export JULIA_NUM_THREADS=100

and then launch julia as usual. Once in julia session, run Threads.nthreads(). This should print a number larger than 1 if your hardware supports multithreading.

In order to quickly test multithreading, do the following to compare the speed of serial and parallel version of myupdate function in src/JuliaTutorial.

julia> using Pkg

julia> Pkg.activate("./")

julia> using JuliaTutorial

julia> y = rand(1_000_000); x = rand(1_000_000);

julia> myupdate!(y, x); myparallelupdate!(y, x); # run them once to initiate compilaoin

julia> @time myupdate!(y, x)

julia> @time myparallelupdate!(y, x)

Try making the size of y and x bigger to see the power of parallelization!