Kick-off script

[joshfactorial]

Give a folder that was output by the splitting script, this script will parse the folder for run files and execute each run file in parallel (multi-threading). As an option, it could just output the command to execute for each run. For example, if the user wanted to submit multiple jobs to a supercomputing cluster, then the script could issue the submission script for each command (with custom header text).

Example 1:

Input header:

#!/bin/bash
#SBATCH --job-name="my_neat_runs_{}"
#SBATCH --output="my_neat_runs_%j.out"
#SBATCH --partition=gpu
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=4
#SBATCH --cpus-per-task=40
#SBATCH --cores-per-socket=20
#SBATCH --threads-per-core=4
#SBATCH --sockets-per-node=2
#SBATCH --mem-per-cpu=1200
#SBATCH --export=ALL
#SBATCH --gres=gpu:v100:4
#SBATCH --time=4:00:00

The program will replace the empty curly braces with a number and generate a run script for each file:

#!/bin/bash
#SBATCH --job-name="my_neat_runs_1"
#SBATCH --output="my_neat_runs_%j.out"
#SBATCH --partition=gpu
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=4
#SBATCH --cpus-per-task=40
#SBATCH --cores-per-socket=20
#SBATCH --threads-per-core=4
#SBATCH --sockets-per-node=2
#SBATCH --mem-per-cpu=1200
#SBATCH --export=ALL
#SBATCH --gres=gpu:v100:4
#SBATCH --time=4:00:00

neat read-simulator -c file1_cfg.yml -o my_specified_output

For input we'll need the original folder, the header template (if generating batch scripts). Optionally, it can just launch the runs in a multi-threaded instance, which will take longer. It should be tied into NEAT's logging features. It would need a list of command line commands to use in the output script.

This may even be two separate scripts: batch script generator, multi-threaded kickoff. But there will be a lot of overlap.