Compiling C/C++

Assuming you already have access to the cluster, you can follow these steps to access your desired compiler, move your file to the HPC, schedule your job and run your code.

Load a Compiler Module

In order to load a module, you can use the 'module load' command followed by the name of the specific compiler you'd like to use. For example, in order to load the default GCC compiler you can use the following command:

To compile, we first need to load our compiler as a module:

module load gcc

Then, simply run

gcc main.c

We get an output file called a.out, which we can run like this:

./a.out

Or, we can name it something, like “hello”:

gcc -o hello main.c

./hello

To compile multiple files, we can run:

gcc file1.c file2.c file3.c

To make compilation easier, we can write a makefile, which is a script that simplifies building for various targets. Makefiles are always named “Makefile”, and they look like this:

  foo:
    gcc -o foo main1.c main2.c
  bar:
    gcc -o bar main3.c main4.c
  
  all: foo bar
  clean:
    rm -f foo bar

Here, our targets are *foo* and *bar*, which we can simply compile like this:

make foo

make bar

If we want to build both at the same time:

make all

By default, make builds all targets, so

make

would build both foo and bar.

To clean up the binaries in the project directory, we can run

make clean

Targets can be combined, like

make clean all

to clean first, then rebuild all targets.

In a SLURM job:

You can run your binary just like any other program on the WAVE HPC using SLURM. Let’s use batch to submit a SLURM job:

First, let’s assume we have a project directory called “project” in the home directory of your WAVE user:

/WAVE/users2/unix/<user>/project

Assuming you have source code on your local machine, let’s compile it first:

make myprogram

Then, we can simply copy over the binary to the cluster:

scp myprogram <user>@login.wave.scu.edu:~/project

Then, we can ssh into the login node, and navigate to the project directory:

ssh <user>@login.wave.scu.edu

cd ~/project

Now, we can edit our batch script:

vim job.sh

Here’s what it might look like:

  #!/bin/bash 
  # 
  #SBATCH --job-name=myprogram 
  #SBATCH --output=myprogram-%j.out 
  # 
  #SBATCH --partition=cmp
  #SBATCH --nodes=1 
  #SBATCH --ntasks=1 
  #SBATCH --cpus-per-task=6 
  #SBATCH --mem-per-cpu=4G
  #SBATCH --time=02:00:00 
  #
  #SBATCH --mail-user=<user>@scu.edu
  #SBATCH --mail-type=END
 
  # run the binary:
  ./myprogram

For further reference on writing batch scripts for SLURM, navigate to the resource scheduling page: https://wiki.wave.scu.edu/doku.php?id=hpc_userguide:hpc_resource_scheduling

Then, we can queue up our job like usual:

sbatch job.sh