Objective#

This tutorial will show you how to run a CP2K simulation using the H20-64 benchmark available on the official CP2K website. This use case simulates a system containing 64 water molecules (192 atoms, 512 electrons) in a 12.4 ų cell, with MD running for 10 steps.

We will also demonstrate Inductiva’s ability to efficiently scale this use case on a more powerful machine.

Prerequisites#

To follow this tutorial, download the input file for the H2O-64 benchmark from here and place it in a folder called H2O-64. Once you have the simulation file, you’re ready to scale your simulations to the Cloud.

Running Your Simulation#

Here is the code required to run a CP2K simulation using the Inductiva API.

In this example, we will execute the simulation on a virtual machine with 16 vCPUs, supported by a 4th generation AMD EPYC™ (Genoa) processor.

"""CP2K Simulation."""
import inductiva

# Allocate a machine on Google Cloud Platform
cloud_machine = inductiva.resources.MachineGroup( 
    provider="GCP",
    machine_type="c3d-highcpu-16")

# Initialize the Simulator
cp2k = inductiva.simulators.CP2K( 
    version="2025.1")

# Run simulation
task = cp2k.run( 
    input_dir="/Path/to/H2O-64",
    sim_config_filename="H2O-64.inp",
    on=cloud_machine)

task.wait()
cloud_machine.terminate()

task.download_outputs()

task.print_summary()

To adapt this script for other CP2K simulations, replace input_dir with the path to your CP2K input files and set the sim_config_filename accordingly.

Once the simulation is complete, we terminate the machine, download the results, and print a summary of the simulation as shown below.

inductiva tasks info 7y33ri1umg0tsppabxf5jska4

Task status: Success

Timeline:
	Waiting for Input         at 05/03, 16:04:17      1.525 s
	In Queue                  at 05/03, 16:04:19      21.308 s
	Preparing to Compute      at 05/03, 16:04:40      8.307 s
	In Progress               at 05/03, 16:04:48      103.301 s
		└> 103.177 s       /opt/openmpi/5.0.6/bin/mpirun --use-hwthread-cpus --np 16 cp2k.psmp H2O-64.inp
	Finalizing                at 05/03, 16:06:32      0.453 s
	Success                   at 05/03, 16:06:32      

Data:
	Size of zipped output:    87.62 KB
	Size of unzipped output:  296.14 KB
	Number of output files:   6

Estimated computation cost (US$): 0.0049 US$

Go to https://console.inductiva.ai/tasks/7y33ri1umg0tsppabxf5jska4 for more details.

The core computation time for this simulation was approximately 1 minute and 43 seconds (103 seconds), as shown in the In Progress line. This represents the actual execution time of the CP2K benchmark on a 16 virtual CPU machine.

For comparison, the same simulation takes 1 minute and 15 seconds on a similar local machine with a 16 virtual CPUs (Ryzen 7 7700X). This performance difference is expected, as cloud CPUs typically have lower clock speeds compared to regular desktop processors, prioritizing energy efficiency and density over raw speed.

However, increasing the number of vCPUs on the cloud machine could enhance performance.

Scaling Up Your Simulation#

With the Inductiva API, scaling up your CP2K simulation is as simple as changing two parameters:

1. Modify the machine_type to a more powerful machine with more vCPUs.

2. Adjust the n_vcpus accordingly to maximize parallel processing efficiency.

As mentioned above, running this simulation on a 16 vCPU cloud machine was slower than on a similarly powered local computer. To enhance performance, we upgraded to a c3d-highcpu-60 instance with 60 vCPUs, reducing the runtime to just 38 seconds—with a slight cost increase to $0.0077.

By leveraging the Inductiva API, you can efficiently scale your CP2K simulations to meet your computational needs. Try different machine configurations and optimize your workflow for faster, more cost-effective results!