CADFEM: Over 300 simulation projects annually in the cloud
Complex calculations: CADFEM simulates physical relationships and creates elaborate models of future products. For this, the company uses flexible resources from T Cloud Public.
In this article, you’ll read
why CADFEM has relied on the cloud for years, both for customer projects and across its other technical departments,
why flexible computing resources (CPUs and GPUs) are ideal for simulation projects
and what other benefits a European cloud offers engineering disciplines
The case at a glance
Challenge:Peak workloads across more than 300 projects per year push local computing resources to their limits.
Solution: CADFEM uses T Cloud Public to flexibly adjust computing resources (CPUs and GPUs) to the needs of its simulation projects.
Result: Computing resources can be tailored to different simulation tasks, allowing CADFEM to process customer projects quickly, regardless of current workload. At the same time, much of the IT administration and capacity planning is eliminated, while infrastructure costs can be allocated to individual projects.
Project facts
Customer
CADFEM Europe
Industry
High-performance computing for engineering requirements
CADFEM Europe and CADFEM APAC form the world’s largest Ansys Channel Partner network under the umbrella of the CADFEM Group, with 600 employees in 20 countries. CADFEM helps customers understand complex physical relationships using numerical simulation and turn the results into practical insights for product development. Depending on the task, the company uses structural mechanics, fluid mechanics, electromagnetics, or coupled physics simulations, among other disciplines. The computing power required can vary considerably depending on the project and simulation method.
CADFEM offers simulation software, consulting, training, and other value-added services, as well as performing simulations on behalf of its customers.
The challenge How can CADFEM obtain sufficient computing resources for HPC simulations?
In the past, CADFEM maintained its own computing power for simulations. CADFEM’s simulation projects differ considerably in their physical requirements and, therefore, in their computing needs and runtimes. While some projects require relatively few resources, complex calculations can require many times more computing power at short notice. As a result, computing power requirements cannot always be predicted well in advance—and bottlenecks sometimes occurred. “When several compute-intensive projects were running at the same time, we had to coordinate the available resources very carefully and prioritize between projects,” says Markus Meingast, Head of Engineering Services at CADFEM. CADFEM did not want to accept this situation. The company wanted to bring its high quality standards into line with the dynamic demands of its project-driven business.
We wanted to free ourselves from the limitations of our own hardware, make capacity management for simulations more flexible, and give employees greater freedom.
Markus Meingast · Head of Engineering Services CADFEM
The solution How can simulation projects be managed more flexibly?
Several years ago, CADFEM established a new approach with T Cloud Public: Instead of maintaining its own computing infrastructure, the company uses computing power as a resource on demand. Today, the cloud is the primary source of computing resources for its simulation projects.
CADFEM Europe now runs over 300 customer projects per year in the cloud. For each simulation task, CADFEM defines the environment required to calculate and evaluate the engineering project. T Cloud Public provides the necessary HPC resources reliably and flexibly. “With the cloud, we now have a new engine room that can easily handle extensive HPC calculations. Whenever necessary, we can increase resources for individual simulations at short notice to reduce simulation times,” Meingast adds.
Easy capacity adjustments with the cloud
For CADFEM, flexibility is one of the key advantages of the cloud model. Whether a simulation needs hundreds of CPU cores at short notice or another project can manage with just a few machines, T Cloud Public gives each project exactly the computing power it needs—without concurrent projects competing for resources. For compute-intensive simulations, hundreds of CPUs and GPUs can be provisioned at short notice when needed, without CADFEM having to maintain the corresponding hardware permanently.
This fundamentally changes day-to-day work. Employees no longer need to reserve resources well in advance or check which machines are available.
CPU or GPU—both from the cloud
Solvers and simulation applications continue to evolve, and so do the demands on computing infrastructure. Depending on the simulation task, software, and solver—for example, in structural or fluid mechanics simulations—CPU- or GPU-based systems may offer advantages. The appropriate architecture for a project is therefore determined by the simulation task. CADFEM can use different computing resources depending on the application, rather than having to adapt the simulation task to a fixed hardware infrastructure.
At CADFEM, approximately 65% of the computing power currently used is CPU-based, while around 35% is GPU-based. The reason: Many simulation applications are still optimized for CPU computing. At the same time, new solvers and GPU-enabled software increasingly make it possible to use graphics processors for demanding calculations. The differences can be considerable. In suitable applications, for example, fluid simulations that can take several days on CPU systems can be calculated in minutes with GPU support.
For the cloud model, this does not represent a fundamental change. On the contrary, this is precisely where it shows its strength: “When the requirements of the simulation software change, we can find the right computing resources in T Cloud Public—whether they require CPUs or GPUs,” says the Head of Engineering Services.
Computing power from Germany
In addition to flexibility, the location of computing resources is important to CADFEM. Customer projects generate valuable intellectual property. The fact that the required computing power is provided in Germany is therefore a key factor in the company’s use of T Cloud Public.
Personal contacts
Another benefit is the collaboration with Deutsche Telekom. CADFEM particularly values the service provided by its personal T Cloud Public contacts—an aspect that, in the company’s view, distinguishes it from large, impersonal cloud platforms. T Cloud Public thus combines flexible cloud computing power with a level of personal connection and reliability that matters to CADFEM.
Why is T Cloud Public an ideal fit for project-based business?
Over 300 customer projects per year, each with different requirements: For CADFEM, the on-demand and largely automated provisioning of computing power is a factor in business success.
Results and key figures
Over 300 customer projects per year: Simulations for customers from various countries and industries
65% CPU / 35% GPU: Flexible computing architecture based on the requirements of the simulation software
10–400 CPUs: High scalability at short notice for individual simulations
Cloud-based for years: T Cloud Public serves as a flexible computing infrastructure for CADFEM’s business
Before and after: A direct comparison
Before: On-premises infrastructure 🔴
After: Flexible cloud resources 🟢
Maintain dedicated machines for simulations
Provision computing resources from the cloud on demand
Plan capacity for expected peak workloads
Adjust capacity to project requirements at short notice
Waiting times occur depending on demand
Start HPC simulations immediately
Fixed machine types and available hardware
Different machines suited to each requirement
Distribute limited resources internally
Two machines today, 10 tomorrow, available at short notice
Administer and maintain dedicated hardware
No in-house effort for operations and hardware
Invest in dedicated computing capacity
Use computing power as a flexible resource
Cloud: A proven delivery model for computing resources
Using the cloud is no longer an experiment for CADFEM. For several years, the company has relied on T Cloud Public as the computing infrastructure for its HPC simulation projects. The cloud has become CADFEM’s new engine room for simulation. Applications range from electromagnetic simulation of semiconductor switching behavior for Osram to the optimization of mixing processes for Liebherr and the simulation of antenna behavior in connected IoT household appliances for BSH. CADFEM also reduced aerodynamic drag for truck manufacturer Krone. These examples demonstrate the diverse applications of modern simulation technologies.
The IT team has less infrastructure to administer, while simulation teams have access to the computing power they need for their customer projects at any time, regardless of project size or complexity. The cloud model also benefits CADFEM’s customers: Cloud transparency allows CADFEM to see exactly how many resources are actually used for a project and allocate infrastructure costs precisely to individual customer projects.
With T Cloud Public, we gain greater freedom, become independent of the limitations of our own hardware, and have cost transparency. Above all, we can respond flexibly to new market requirements and future customer needs.
Markus Meingast · Head of Engineering Services CADFEM
T Cloud Public services used
Elastic Cloud Server — CADFEM uses General Purpose, Memory Optimized, and GPU Accelerated flavors. “We’re seeing a clear performance gain with the new-generation CPUs (c9/m9),” confirms Meingast.
Object Storage Service — CADFEM stores more than 7 TB and up to 12 TB of simulation data in object storage.
Elastic Volume Service — CADFEM flexibly stores 1–7 TB of data in different block storage performance classes.
Frequently asked questions
Is the cloud suitable for simulations?
Simulations are a prime example of workloads that require substantial resources for a limited time. The cloud is an ideal solution for these workloads, as well as other high-performance computing scenarios, in terms of both cost and performance.
What benefits does the cloud offer for HPC?
The cloud eliminates the need for in-house capacity planning and IT administration. Infrastructure cost transparency enables precise billing to users. GPU resources are also readily and directly available.
Does CADFEM use the cloud as a SaaS platform for its customers?
CADFEM primarily uses T Cloud Public to carry out simulation projects on behalf of its customers. Customers can now also use Simulation as a Service through a self-service model, with direct access to the required CPU and GPU resources in various predefined capacity sizes.
Why does CADFEM need both CPUs and GPUs?
The most suitable hardware depends on the simulation software and the solvers used. Today, approximately 65% of CADFEM’s usage is CPU-based and 35% is GPU-based. As GPU support in simulation software increases, this ratio may continue to change.
What opportunities does the cloud offer simulation providers?
As a next step, simulation providers such as CADFEM can build their own web portals, allowing customers to run simulations on a self-service basis. Powerful cloud resources operate “invisibly” in the background and can be allocated using features such as auto scaling.
Why is the cloud location important to CADFEM?
Simulation projects can involve the processing of sensitive customer data and generate unique insights that provide a competitive advantage. Processing data in Germany is therefore an important factor for CADFEM in protecting its customers’ intellectual property.
How many computing resources can CADFEM use for a simulation?
Depending on the project, several hundred CPUs can be provisioned at short notice. What matters is not having permanent capacity available, but provisioning resources on demand for each project.
Is T Cloud Public the right fit for your use case?
Talk with a cloud expert at no cost and with no obligation.
Request a Consultation Now
This content might also interest you
Copernicus Data Space Ecosystem
Copernicus Data Space Ecosystem of the European Space Agency (ESA)
More about Copernicus
Bosch Subsidiary relies on the T Cloud Public for product protection
With Origify, Bosch is revolutionizing product authentication
Learn more about Origify
Fuse-AI: An algorithm for detecting tumors
Fuse-AI develops artificial intelligence for cancer detection
More about Fuse-AI
T Cloud Public Community
This is where users, developers and product owners meet to help each other, share knowledge and discuss.
Discover now
Free expert hotline
Our certified cloud experts provide you with personal service free of charge.