added slurm scripts and a readme for using them

slurm/README.md

@@ -0,0 +1,13 @@
+## Slurm instructions
+
+the `srun.sh` script is a standalone script to run a single slurm job in the current working directory. This should usually be run in its own directory as it will generate a number of outfiles, as well as data and job id files.
+
+it can be run, once the desired slurm and python parameters for the job are set in `srun.sh`, using:
+
+    sbatch srun.sh
+
+The `sbatch.sh` file is used to batch creat a number of slurm jobs, and will generate its own measurement subdirectories for each parameter set. The example files will run 32 measurements on 32 cores for each job, and a job will be submitted for 6 stick densities, from 4 to 20 per um.
+
+slurm parameters are set in the `srun_header.sh` file, while parameters for the python job are set in `sbatch.sh`. These must be checked to ensure they match, and for that reason could be integrated into a single script at some point. Usage is:
+
+    bash sbatch.sh

slurm/sbatch.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+# this dictates the number of measurements to be made, and for what densities
+for density in 4 9 10 11 14 20
+do
+    n=$(echo $density*3600 | bc)
+    echo $n
+    mkdir meas_32x$density
+    cp srun_header.sh meas_32x$density/srun.sh
+    echo 'python3 ~/gitrepos/networksim-cntfet/measure_perc.py -s --cores 32 --start '$n' --step 0 --number 32 --scaling 60'>>meas_32x$density/srun.sh
+    cd meas_32x$density/
+    output=$(sbatch srun.sh)
+    echo $output
+    touch 'jobid_'${output:20}
+    cd ..
+done

srun.sh → slurm/srun.sh

@@ -4,6 +4,6 @@
 #SBATCH -o outfile  # send stdout to outfile
 #SBATCH -e errfile  # send stderr to errfile
 #SBATCH -t 10:00:00  # time requested in hour:minute:second
-#SBATCH --mem-per-cpu=1024
+#SBATCH --mem-per-cpu=1024 #memory per task in MB
 #SBATCH --mail-user=$USER@localhost
 python3 ~/gitrepos/networksim-cntfet/measure_perc.py -s --cores 32 --start 36000 --step 0 --number 32 --scaling 60

slurm/srun_header.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+#SBATCH -N 1      # nodes requested
+#SBATCH -n 32     # tasks requested. default is one core per task
+#SBATCH -o outfile  # send stdout to outfile
+#SBATCH -e errfile  # send stderr to errfile
+#SBATCH --mem-per-cpu=1024
+#SBATCH -t 24:00:00  # time requested in hour:minute:second
+#SBATCH --mail-user=$USER@localhost