VSC quick start

Vienna Scientific Cluster user documentation: https://wiki.vsc.ac.at

Introduction to working on the VSC-3 Cluster: https://service.vsc.ac.at/slides/introduction-to-vsc/01_supercomputers_for_beginners/00_intro.html#(3)

Quick start with this project on VSC-3 (for more or updated information read documentation linked above):

1. Clone this github repository to your VSC folder
2. If needed, make changes in Makefile
3. Load modules needed for building a binary file (can be adapted): module load gcc/7.3.0 cmake/3.9.6
4. Build with ./build.sh
5. Prepare or copy data into your local repository (see points 2-7 in YOLOv3 training)
6. Create a slurm job with a chosen darknet command (see points 8-12 in YOLOv3 training). An example for running on one gpu, limited to 3 hours (optional but can shorten the time of waiting for starting the job): retour_train
7. Send the job with sbatch command e.g. sbatch batches/retour_train
8. You can see your jobs with squeue -u yourusername
9. As an output see the most recent .out file