GPU-Accelerated Simulations on Inductiva.AI – Faster, More Efficient HPC with Inductiva API v0.14

The Inductiva Team

Author

February 17, 2025

Tags:

CFD SimulationsCloud machine allocationCoastal DynamicsCoastal EngineeringComputational DesignDocker and ApptainerHow to choose the best cloud machine for simulationsNumerical SimulationsInductiva API v0.14 releaseSNL-SWANCPK2GXGPUGROMACS
Inductiva API v0.14 release

Introducing Inductiva API v0.14 – a powerful upgrade packed with GPU-powered numerical simulations, real-time monitoring of simulation outputs, seamless AWS exports, and an expanded portfolio of built-in simulators. This release takes high-performance computing (HPC) to the next level, offering greater control, efficiency, and flexibility for scientific research in molecular dynamics, coastal dynamics, fusion energy.

With GPU acceleration now available, users can leverage parallel computing for faster, more efficient large-scale simulations—significantly reducing runtime and enhancing performance for demanding computational tasks like CFD, molecular dynamics, and AI-driven physics modeling

Real-Time Monitoring of Output Files for Simulations

Tracking simulation outputs in real time just got better! Previously, users could monitor stdout and stderr streams live (see Stream logs of a task), but other output files were only accessible after task completion. 

Now, with v0.14, all output files can be monitored in real-time while the simulation is running. This lets users debug, track specific variables, and inspect partial results to ensure simulations are running as expected.

New CLI Commands for Real-Time File Monitoring

🔍 List all output files generated by a task:

		inductiva tasks list -files <task_id>
	

📌 Tail any non-binary output file in real-time:

		inductiva tasks tail <task_id> <filepath> -f
	

More details are available in our documentation on accessing output files.

GPU-Powered Numerical Simulations on GCP

Some of our supported simulators—like GROMACS (molecular dynamics) and GX (fusion energy simulations)—leverage GPU acceleration for significant performance boosts. Now, you can launch GPU-powered machines on Google Cloud Platform (GCP) to take full advantage of parallel computing for high-performance simulations.

How to Launch a GPU-Powered Machine in Inductiva

One of the most valued benefits of Inductiva is how effortlessly users can select and allocate a machine to run their tasks. Launching a GPU-powered instance is as seamless as starting any other x86 machine—simply choose a machine from the G2 series in the Console and run it with the same straightforward command:


 gpu_cloud_machine = inductiva.resources.MachineGroup(
  provider="GCP",
  machine_type="g2-standard-4" )

🚀 Performance Boost: GPU-enabled instances drastically speed up computation and reduce total runtime, making large-scale simulations even more efficient. This is particularly beneficial for molecular dynamics, CFD simulations, and AI-driven physics modeling.

For a complete example, check out our GX on GPUs tutorial.

Seamless Exporting of Simulation Files to AWS S3 Buckets

All files generated by the simulations are securely stored in the user’s own cloud bucket, managed by Inductiva on Google Cloud Platform. 

To streamline workflows involving other clouds, users are now able to export simulation results from their Inductiva bucket to an AWS S3 bucket directly, thus avoiding the manual and time-consuming task of downloading to local computers and uploading to AWS.

New CLI Command for AWS S3 Export

💡 Quick Setup: Before using this feature, users need to authenticate with AWS credentials via:

		aws configure
	

More details can be found in our documentation on exporting files to AWS.

This feature to move files directly from Inductiva’s bucket to another cloud is only available for AWS for now, but get in touch with us to let us know which other clouds you work with and how we can streamline your workflow.

New Open-Source Simulators Now Available

Inductiva v0.14 adds support for four new open-source simulators, expanding computational capabilities across various scientific fields:

  • 🌊 SNL-SWAN – Advanced coastal dynamics simulations for ocean engineering.
  • 🔬 CPK2 – Enhancing molecular dynamics simulations (alongside GROMACS).
  • GX – Our first fusion energy systems simulator for plasma physics.

Each simulator is always made available in Kutu, Inductiva’s free repository of Docker images for scientific computing software. The full list of integrated simulators and their supported versions can be found on our website or by running:

		inductiva simulators list
	

Setting and starting a cloud machine streamlined

The simulation process just became even more seamless! Previously, two steps were required to start a cloud machine:

  1. Allocate the MachineGroup;
  2. Start it.

Now, the MachineGroup is automatically started when the task is run:

machine_group = inductiva.resources.MachineGroup("c2d-highcpu-112")
reef3d = inductiva.simulators.REEF3D()
task = reef3d.run(
    input_dir="/path/to/my/reef3d/files",
    on=cloud_machine)

This simple yet powerful improvement is much more intuitive, requiring less commands; it’s also less prone to errors, because the user doesn’t have to explicitly add the command; and more cost effective, because the machine isn’t started until it’s actually needed to run the task.

What’s Next?

At Inductiva, we’re committed to making high-performance computing more accessible and pushing the boundaries of GPU-powered simulations and AWS cloud integration.

📌 Upcoming Features & Improvements:

Usability Enhancements for Simulations & Cloud Computing

  • Huge improvements in storage management
  • Email alerts for task completion 
  • Improved Web Console experience

Team Collaboration

  • Share machine groups with colleagues

Stay tuned—big things are coming for AWS integration, GPU-powered simulations, and HPC cloud computing!

Upgrade Now

Make sure you’re using the latest version of Inductiva by running:

		pip install –upgrade inductiva
	

A heartfelt thank you to our incredible team and community for being so involved and for helping us grow, one simulation at a time!

🎉 Happy Simulating! 🎉

Check out our blog

Blog post banner, blue abstract background

From Supercomputer to Cloud: A New Era for OpenFOAM Simulations

Inductiva joined the 1st OpenFOAM HPC Challenge to test how cloud infrastructure stacks up against traditional HPC for large-scale CFD simulations. Running the DrivAer automotive benchmark, the team explored multiple hardware setups, hyperthreading choices, and domain decomposition strategies. The results? Inductiva’s flexible MPI clusters handled up to 768 partitions with impressive price-performance—even outperforming pricier hardware in some cases. For simulations below massive supercomputer scales, cloud HPC proves not only competitive but cost-effective, offering engineers and researchers agility without sacrificing speed. Curious how to fine-tune your OpenFOAM workloads in the cloud? Dive into the benchmarks and see what’s possible.

Future of AI blog post banner

The Future of AI is Physical: Simulation is Key

Everyone’s talking about LLMs, but the real next wave is AI for the physical world—AI that understands fluid dynamics, thermodynamics, electromagnetism, and even quantum physics. From breakthroughs like DeepMind’s AlphaFold in protein folding to NVIDIA’s AI weather forecasting, industries from renewable energy to materials science are being transformed. But experimental data isn’t always enough; numerical simulations are crucial to generate high-quality datasets that power scientific AI. At Inductiva.AI, we’re offering a computational platform to democratize physics dataset generation, combining open-source simulators with cloud HPC, making it as easy as running a Python script. Explore how AI and scientific computing are converging—and what it means for the future of innovation.

Inductiva API v0.16 release

🔄 Real time Outputs, 💼 Smarter Projects, 🏷️ Metadata at scale, 🔒Private Simulators and two new ways to model the ocean 🌊 : What’s New in Inductiva.AI v0.16

We’re happy to share with our community Inductiva.AI v0.16, a powerful new release designed to help you organize simulations at scale, preserve context with metadata, and gain full control over your simulation environments. This version brings major upgrades to Project management, adds built-in metadata support, and now allows you to run your own private simulators securely. Plus, we’ve expanded our simulator catalog with two new engines for hydrodynamics and coastal modeling: OpenTelemac and Delft3D. Whether you're managing 10 or 10,000 simulations, this release helps you move faster, stay organized, and simulate anything — from river flows to molecular dynamics — on your terms.