Checking which modules are available on the cluster:
module avail
1- Load GCC/11.2.0
module load GCC/11.2.0
2- Load CMake/3.12.1
module load CMake/3.12.1
3- Upload TreeWidzard's zip file to the cluster
scp TreeWidzard.zip <username>@saga.sigma2.no:
Observation: Keep : after saga.sigma2.no.
4- Unzip TreeWidzard's zip file
unzip TreeWidzard.zip
5- Enter the directory of TreeWidzard
cd TreeWidzard
6- Create a directory called Build
NOTE: If the directory exists, remove all content of the directory
7- Enter Build
cd Build
8- Run cmake in Build
cmake ..
9- Run make in Build
make
The program ready to use.
For Test:
In Build folder, run the following command
./treewidzard -atp pw = 3 -pl -nthreads 3 ParallelBreadthFirstSearch ../small-tests/conjecture.txt
This command will test TreeWidzard on a conjecture on graphs of pathwidth at most 3 and using 3 threads.
Make sure treewidzard is working.
Let PROGRAM_FILE be TreeWidzard's executable file.
Let CLUSTER_OUTPUT_PATH be the path that you want to store the cluster outputs.
Let OUTPUT be the output file of TreeWidzard.
Let CONJECTURE_FILE be the conjecture file.
Let SEARCHSTRATEGY be the search strategy.
Example:
#SBATCH --account=nn9535k --job-name=MyJob
#SBATCH --partition=normal
#SBATCH --time=0-00:10:00
#SBATCH --ntasks=1 --cpus-per-task=32
#SBATCH --mem-per-cpu=3G
#SBATCH --output=/CLUSTER_OUTPUT_PATH/slurm_%x_%j.out
#SBATCH --error=/CLUSTER_OUTPUT_PATH/slurm_%x_%j.err
PROGRAM_FILE -atp pw = 3 -pl -nthreads 3 SEARCHSTRATEGY CONJECTURE_FILE > OUTPUT
exit 0
#SBATCH --account=nn9535k --job-name=MyJob
#SBATCH --partition=bigmem
#SBATCH --time=1-0:0:0
#SBATCH --ntasks=1 --cpus-per-task=32
#SBATCH --mem-per-cpu=32G
#SBATCH --output=/CLUSTER_OUTPUT_PATH/slurm_%x_%j.out
#SBATCH --error=/CLUSTER_OUTPUT_PATH/slurm_%x_%j.err
PROGRAM_FILE -atp pw = 3 -pl -nthreads 60 SEARCHSTRATEGY CONJECTURE_FILE > OUTPUT
exit 0