Why wait for cluster time when others are publishing? Run your simulations now and stay ahead.
This guide demonstrates how to run a simulation using a custom Docker image.
By the end, you'll be able to use the CustomImage simulator using your own public or private image.
CustomImage SimulatorThe CustomImage enables you to run simulations using any Docker image of your choice by specifying the container_image parameter.
This gives you full control over the simulation environment.
In addition to container_image, you can provide a list of commands to execute within the container. These are the only parameters unique
to the CustomImage simulator; all other configuration options work just like those for the simulators integrated into our platform.
To illustrate, we’ll use a Docker image of the Fire Dynamics Simulator (FDS), which is publicly available in our Docker Hub repository,
Kutu. Below is an example of how to run a simulation using the CustomImage simulator:
import inductiva
# Allocate cloud machine
machine_group = inductiva.resources.MachineGroup("c2d-highcpu-4")
machine_group.start()
input_dir = inductiva.utils.download_from_url(
"https://storage.googleapis.com/inductiva-api-demo-files/fds-input-example.zip", unzip=True)
custom_simulator = inductiva.simulators.CustomImage(container_image="docker://inductiva/kutu:fds_v6.8")
task = custom_simulator.run(input_dir=input_dir, commands=["fds mccaffrey.fds"],
on=machine_group)
task.wait()
task.download_outputs()
machine_group.terminate()
To adapt this script for your own custom image, simply update the container_image and adjust the commands as needed for your simulation.
A list of commands can include Command instances, each optionally associated with an MPIConfig. Below is an example demonstrating this:
import inductiva
# Instantiate machine group
machine_group = inductiva.resources.MachineGroup("c2-standard-4")
machine_group.start()
input_dir = inductiva.utils.download_from_url(
"https://storage.googleapis.com/inductiva-api-demo-files/fds-input-example.zip", unzip=True)
custom_simulator = inductiva.simulators.CustomImage(container_image="docker://inductiva/kutu:fds_v6.8")
mpi_config = inductiva.commands.MPIConfig("4.1.6", np=4, use_hwthread_cpus=True)
command = inductiva.commands.Command("fds mccaffrey.fds", mpi_config=mpi_config)
task = custom_simulator.run(input_dir=input_dir, commands=[command],
on=machine_group)
task.wait()
task.download_outputs()
machine_group.terminate()
This example runs four instances of your image using the command fds mccaffrey.fds, with MPI version 4.1.6. This is equivalent to:
mpirun -np 4 -use_hwthread_cpus apptainer run ... fds mccaffrey.fds
Alternatively, you can run MPI directly within your container as shown below:
import inductiva
# Instantiate machine group
machine_group = inductiva.resources.MachineGroup("c2-standard-4")
machine_group.start()
input_dir = inductiva.utils.download_from_url(
"https://storage.googleapis.com/inductiva-api-demo-files/fds-input-example.zip", unzip=True)
custom_simulator = inductiva.simulators.CustomImage(container_image="docker://inductiva/kutu:fds_v6.8")
command = inductiva.commands.Command("mpirun -np 4 --use-hwthread-cpus fds mccaffrey.fds")
task = custom_simulator.run(input_dir=input_dir, commands=[command],
on=machine_group)
task.wait()
task.download_outputs()
machine_group.terminate()
This is equivalent to:
apptainer run ... mpirun -np 4 -use_hwthread_cpus fds mccaffrey.fds
MPIConfig. For example, if your container
has OpenMPI 1.10.6, select a compatible version (such as 1.10.7) in MPIConfig.