#!/bin/bash

# SPDX-FileCopyrightText: 2026 Samudra Authors
#
# SPDX-License-Identifier: Apache-2.0

#SBATCH --job-name=%j-samudra-perlmutter-om4
#SBATCH --output=train/logs/slurm-%j.out
#SBATCH --error=train/logs/slurm-%j.err
#SBATCH --account=m4874
#SBATCH --constraint=gpu&hbm80g
#SBATCH --qos=debug_preempt
#SBATCH --mem=110G
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=2
#SBATCH --cpus-per-task=32
#SBATCH --gpus-per-task=1
#SBATCH --gpu-bind=none     # This ensures all GPUs are visible to all tasks
#SBATCH --time=00:20:00
#SBATCH --signal=USR1@60
#SBATCH --requeue
#SBATCH --open-mode=append
srun ./scripts/variable-time-payload.sh
sleep 120 # make sure a process is still running for slurm to send SIGKILL to
