Production-ready templates and curated community tools.

#!/bin/bash
#SBATCH --job-name=my_job
#SBATCH --output=output_%j.log
#SBATCH --error=error_%j.log
#SBATCH --time=01:00:00
#SBATCH --partition=compute
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=4
#SBATCH --mem=16G
 
# Load required modules
module load python/3.11
 
# Run your application
python train.py --epochs 100

#!/bin/bash
#SBATCH --job-name=gpu_training
#SBATCH --output=gpu_%j.log
#SBATCH --partition=gpu
#SBATCH --gres=gpu:a100:2
#SBATCH --time=24:00:00
#SBATCH --mem=64G
 
# GPU setup
export CUDA_VISIBLE_DEVICES=0,1
 
# Load CUDA modules
module load cuda/12.1
module load cudnn/8.9
 
# Run GPU training
python train.py --gpus 2

#!/usr/bin/env python3
import subprocess
import time
import datetime
 
def get_gpu_stats():
    result = subprocess.run([
        'nvidia-smi',
        '--query-gpu=index,utilization.gpu,memory.used,temperature.gpu',
        '--format=csv,noheader,nounits'
    ], capture_output=True, text=True)
    
    for line in result.stdout.strip().split('\\n'):
        idx, util, mem, temp = line.split(',')
        print(f"[{datetime.datetime.now()}] GPU{idx}: {util}% util, {mem}MB mem, {temp}°C")
 
while True:
    get_gpu_stats()
    time.sleep(10)

#!/bin/bash
#SBATCH --job-name=array_job
#SBATCH --output=output_%A_%a.log
#SBATCH --array=1-100
#SBATCH --time=00:30:00
#SBATCH --mem=4G
 
# Process task ID
echo "Processing task $SLURM_ARRAY_TASK_ID"
 
# Run with task-specific parameter
python process.py --id $SLURM_ARRAY_TASK_ID

# CPU Partition
PartitionName=compute Nodes=node[01-20] Default=YES MaxTime=7-00:00:00 State=UP
 
# GPU Partition
PartitionName=gpu Nodes=gpu[01-08] MaxTime=3-00:00:00 State=UP
 
# High Memory Partition
PartitionName=highmem Nodes=mem[01-04] MaxTime=2-00:00:00 State=UP
 
# QOS Configuration
QOS=normal Priority=50 GrpTRES=cpu=1000
QOS=high Priority=100 GrpTRES=cpu=200