jaantollander / cell-lists-benchmarks

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Cell Lists Benchmarks

Scripts for running CellListsBenchmarks.jl from the command line and plotting the results.

The benchmarks/run.jl file contains the script to run benchmarks from the command line. We can test the commands locally with benchmarks/test.sh. The test script also demonstrates how to use the commands.

Usage on Aalto Triton Cluster

First, lets connect to Aalto VPN.

sudo openconnect https://vpn.aalto.fi

Then, we can connect to Triton using SSH.

ssh USERNAME@triton.aalto.fi

Let's navigate to the work directory and clone the repository.

cd $WRKDIR
git clone https://github.com/jaantollander/cell-lists-benchmarks.git

Then, we can navigate the repository directory and load the Julia module.

cd cell-lists-benchmarks
module load julia

We need to install CellLists.jl and CellListsBenchmarks.jl manually from GitHub and then activate the environment to install other dependencies from Project.toml. Open the Julia REPL by typing julia the command line.

julia> ]
pkg> add https://github.com/jaantollander/CellLists.jl
pkg> add https://github.com/jaantollander/CellListsBenchmarks.jl
pkg> add ArgParse Dates Random JLD

We should start by changing our working directory to benchmarks and create the output directory.

cd benchmarks
mkdir output

Then, we can submit a batch jobs to the Slurm scheduler with the sbatch command.

sbatch brute_force.sh
sbatch cell_lists.sh
sbatch cell_list_constructor_serial.sh
sbatch cell_list_constructor_threads2.sh
sbatch near_neighbors_serial.sh
sbatch near_neighbors_threads4.sh

We can access the benchmark data on Triton by remote mount the work directory to our local computer. We can remote mount our work directory using sshfs.

mkdir ~/triton_work
sshfs USERNAME@triton.aalto.fi:/scratch/work/USERNAME ~/triton_work

About

License:MIT License


Languages

Language:Julia 67.1%Language:Shell 32.9%