Engineering simulation is used across industries to accelerate product development. Simulations are used to check the safety of aircraft, cars, and buildings, ensure that your mobile phone has a signal wherever you go, and maximize the range of the newest electric vehicles. It reduces the need for expensive, time-consuming physical testing and enables engineers to iteratively improve designs much faster.
Unfortunately, running complex simulations can also be time-intensive and requires significant high-performance computing (HPC) resources.
Luminary Cloud, a member of the NVIDIA Inception program for startups, has been developed from the ground up to take advantage of the latest cloud and NVIDIA GPU technologies, removing the computational burden from organizations and accelerating these simulations.
In this post, I discuss the challenges to a wider adoption of simulation in industries and how Luminary Cloud uses the latest cloud and NVIDIA GPU-accelerated computing to address these challenges. I also discuss a real-world case study with Joby Aviation.
Challenges in computational fluid dynamics
In the realm of computational fluid dynamics (CFD), where intricate simulations shape the future of engineering and scientific advancements, several challenges persist, hindering widespread adoption.
A recent survey highlights these challenges:
- Simulation turnaround time: Typical CFD projects can take a significant amount of time: weeks (36.2%), months (22.3%), or continuous (24.5%).
- Simulation robustness: Key improvements desired in CFD software include simulation robustness, automatic meshing capabilities, accuracy, and a good price-performance ratio.
- Model preparation: In some cases, setting up CFD simulations can take months. Inefficient model preparation is a significant challenge (44%).
- Cost: 50% of non-users cite high license costs as the main barrier to adopting CFD software. Budget constraints are the most challenging factor for switching software.
Accelerated simulation
The accelerated processing capabilities of GPUs have unlocked new horizons in quantum computing, climate science, financial engineering, and artificial intelligence. Their far-reaching applications enable businesses to achieve processing speeds up to 1K times faster than CPU-only processing.
CFD is an area that has particularly benefited from the accelerated computing capabilities of NVIDIA GPUs, with accelerations of over 30x compared to traditional CPU-based compute. For more information, see The Computational Fluid Dynamics Revolution Driven by GPU Acceleration.
HPC in the cloud
One of the challenges for CFD is access to HPC resources. The compute resources to support an engineering development program are rarely constant. During the design cycle, the need for simulation will peak and drop between phases and programs. This results in inefficiencies of compute capacity, which can remain idle between phases and delay simulation work that is waiting for resources during design phases.
The cloud increases speed and agility as distributed users can easily access powerful HPC resources when they’re needed. Leveraging the elasticity of the cloud also enables companies to right-size their HPC resources day-to-day, increasing cost efficiency and reducing time-to-market.
Cloud resources also enable companies to take advantage of the latest hardware, such as NVIDIA GPUs, which might not be available in-house. A SaaS approach enables increased collaboration, so multiple users distributed globally can access simulations without the need for inefficient data transfer, synchronization, or individually dedicated licenses.
Luminary Cloud’s solution
Luminary Cloud is a multi-physics solution that currently supports CFD for fluid-flow physics and conjugate heat transfer (CHT) for thermal management. The company has addressed the challenges mentioned above simultaneously by developing a novel computer-aided engineering (CAE) tool. Its cloud-based SaaS platform is poised to redefine the landscape of computational engineering, offering a first-of-its-kind, near-real-time engineering experience. Much like using collaborative document editing software, Luminary Cloud’s platform provides a seamlessly interactive platform for customers.
Lumi AI, Luminary Cloud’s AI-based engineering design copilot, cuts down the time that engineers spend in setup and simulation so that they can spend more time on analysis and optimization. For example, Lumi Mesh Adaptation replaces the traditionally tedious steps of mesh generation by intelligently adapting the computational mesh for higher accuracy and efficiency.


Luminary Cloud built their CAE platform specifically to take advantage of multi-node NVIDIA GPU accelerated computing. Other CAE tools were built for multi-node CPU clusters. A subset of those is being ported to use GPU acceleration today.
So, what happens when GPU and cloud computing technological revolutions come together? They form a powerful combination whose value is more than the sum of its parts. By introducing GPU allocation into widely used cloud services, such as Google Cloud, your business can enhance workflow efficiency for products while benefiting from adaptable, pay-as-you-go pricing that can be tailored for unique budgets.

One standout feature is the lightning-fast simulations that run in parallel (both multi-GPU and multi-node), drastically reducing wait times and enabling engineers to focus on what truly matters: achieving design goals within time and cost constraints. Luminary Cloud has seen speed improvements of over 100x compared to traditional approaches.
In a demo with Luminary Cloud, they ran 20 simulations in parallel, each of which used meshes with 40M control volumes (CVs) and was completed in less than two minutes using a modest number of GPUs. Luminary Cloud also ran a steady-state Reynolds number averaged Navier-Stokes (RANS) calculation on a full aircraft geometry during take-off with 150M CVs in about seven minutes using 40 NVIDIA A100 GPUs! The same calculation would have taken 2–3K cores between 6–8 hours.
This leap forward in throughput is further enhanced by ease of use in the form of the tool’s automatic GPU allocation, ensuring that resources are optimized for each task. However, you also have the flexibility to manually select GPU allocation if you have specific requirements or preferences.
Case study: Joby Aviation

Joby Aviation offers a real-world example of the combined power of Luminary Cloud’s platform and NVIDIA GPUs. The Luminary Cloud platform has enabled Joby to quickly assess new designs and pursue only the most promising concepts.
“You can take complete aircraft configurations and run them in a matter of minutes. It allows a level of confidence that was unprecedented before. You can quickly say ‘Yes, this will or will not work’,” explained Joby chief aerodynamicist Gregor Mikić.
The computational power of NVIDIA GPUs plays a key part in getting insights faster, but Luminary Cloud’s suite of front and backend tools also accelerates pre– and post-processing.
For example, Joby Aviation had to iterate rapidly on an accessory part for the aircraft. Speed is essential for risk management within the certification process. A month-long wait for an essential redesign could result in delayed regulatory approval and, ultimately, revenue. Despite the urgency, a project like this would typically require several months using a legacy CFD tool, due to intricate pre-processing steps and long compute times.
Fortunately, by using Luminary Cloud, Joby Aviation was able to complete the task with 10x greater engineer productivity.
The potential rewards for being first to market with an electric air taxi are significant, and Luminary Cloud is proud to support Joby as they develop and introduce this technology.
Cloud-based CAE benefits
Luminary Cloud’s solution should be on the radar of anyone in the CAE field. It provides a novel, real-time engineering experience, facilitating highly interactive simulations that significantly expedite engineering processes. Luminary Cloud also provides a range of modeling approaches from RANS to wall-modeled, large eddy simulation (WMLES), so that you can select the appropriate accuracy level for your project’s time constraints.
Luminary Cloud’s system is designed for faster parallel simulations, reducing wait times and enabling engineers to efficiently tackle complex tasks. Its user-friendly interface, reminiscent of Google Docs, focuses on simplicity and ease of use, ensuring a distraction-free experience.
By choosing Luminary Cloud, you can not only enhance productivity but also explore new frontiers in computational engineering, all powered by the latest technology available in Google Cloud. With the robust combination of NVIDIA GPUs and an array of benefits offered by Google Cloud, Luminary Cloud offers a unique and powerful option for computational engineering.
To learn more about Luminary Cloud, see the High-Performance Computing for Realtime Engineering solution brief. You can also dive deeper into accelerated computing in the NVIDIA Developer Forums.
Learn how the NVIDIA Inception Program helps startups innovate and grow with free benefits and support.