For companies focused on manufacturing areas demanding heavy engineering simulations, unbridled compute power, or the management of data-intensive applications and workflows, challenges multiply quickly. They need a simulation platform that effectively solves these challenges, providing a path to accessing HPC in the cloud to stimulate innovation.
Within the last few years, the cloud has evolved to be more accommodating for engineering workloads thanks to ever-increasing compute power available via faster and cheaper hardware and the maturity of manufacturing software solutions. Early adopters implemented a cloud-first strategy on the enterprise side of the business, leaving HPC lagging. Today, the benefits of the cloud that the enterprise side enjoys are finally available for manufacturing applications.
For HPC in the cloud to work well, several challenges stand in the way.
Acquisition costs associated with owning and operating an on-premises HPC system often represent too great a burden for companies to shoulder, costing millions. Steering through a cumbersome process, companies may wait six months or more before selecting a system, which must be shipped and installed. All previous system and application software must be moved onto the new system. Getting a simulation up and running necessitates porting the latest release over to the new system (typically three times / year), taking several days per instance.
Consider the total cost of ownership (TCO). Costs include hardware as well as the software running on it, the space where hardware is kept, power usage, personnel, and more. TCO calculations must also account for organizational, operational, and logistical costs, demanding ongoing effort and investment to keep a system operating.
Another challenge — a company buys an HPC system today, only to realize that tomorrow, it’s already aged. Running expensive simulation software on an outdated system can take much more time, further increasing TCO.
Enter UberCloud: Seamlessly integrated HPC in the cloud
UberCloud’s fully automated, self-service, multi-cloud engineering simulation platform and standard simulation workflow containers enable enterprises in manufacturing to seamlessly integrate HPC into a “cloud-first” or hybrid cloud strategy. Applying the latest application — and resource-agnostic IT technologies like containers and Kubernetes — UberCloud enables performance without compromise. Engineers can run their complex simulation workflows from anywhere, including their home office, in a fast and familiar way, always using the latest technology.
Widely applicable to a variety of complex undertakings that include multi-physics simulations, construction, AI / machine learning, digital twins, data analytics, autonomous driving, and edge computing, the UberCloud platform and its application containers deliver the compute power, flexibility, and agility required by the most compute- and data-intensive applications across many industries. Because of its capability in making HPC access and use transparent, the UberCloud platform accelerates the recent trend of “democratization of HPC” —it enables more engineers to apply HPC to solve challenging problems.
UberCloud and Rimac Technology
Running computer simulations in the cloud accelerates innovative, high-quality development, optimizes agility, and speeds time to market.
For Rimac Technology, accelerating their research-and-development pipeline to design and deliver powertrains, batteries, and the latest electric-vehicle technology would make them more competitive, innovative, and cost-efficient. The first step required an evaluation of their aging on-premises HPC Cluster and the possibility of moving engineering simulation workloads to the cloud. UberCloud and the Microsoft Azure project team worked with Rimac’s experts to design and configure their Azure cloud environment for running multiple computer-aided engineering (CAE) software calculations to benchmark applications, analyze TCO for cloud versus on premises, and better understand engineering simulation usage of cloud HPC resources and related workflows.
UberCloud engineering simulation platform and container technology, running in Microsoft Azure and using 3rd Gen AMD EPYC ™ processors, integrates easily with existing IT, enabling engineers to run challenging simulations, test larger batches of prototypes faster than ever, and do so securely, driving innovation . Rimac engineers now have on-demand access to additional and more powerful computing resources in the cloud, without having to leave the comfort of their workstations. They gain agility through shorter product design cycles, the direct result of shorter simulation times. They gain a superior level of quality by simulating more sophisticated geometries and physics, and by simply running more iterations. Because it’s the cloud, Rimac pays only for what it uses, further enabling rapid innovation.
UberCloud created a flexible global HPC architecture, able to handle various workloads, which can easily expand to other sites and provide maximum flexibility and cost-efficiency using on-demand, reserved, and spot instances. Knowing that one of Rimac’s key technologies is high-performance battery packs, the UberCloud team assisted Rimac engineers to optimize designs with numerical simulations analyzing mechanical integrity, thermal behavior, and electric current distribution. In the end, HPC on Azure with UberCloud enabled Rimac to significantly improve their overall design processes and increase simulation throughput.
Rimac engineering workflows were analyzed to provide the best cloud environment for the engineering processes, reducing IT management overhead. Engineers can use simulation applications and a vast array of resources on demand, allowing faster results and turnaround time, while simultaneously providing complete control and management to the team. With Azure, engineering simulations gain scalable infrastructure, while data stored in Azure’s secure data centers eliminates any data transfer delays.
What’s next: UberCloud and the manufacturing industry
UberCloud and Microsoft’s work with Rimac and other partners can be applied to any manufacturing design scenario. From machine learning to design of experiments, parameter space evaluation, optimization algorithms, and more, UberCloud’s technology is horizontal — it’s all about engineering, scientific simulations, and data processing. Together, these resources let organizations move from running 10 to 50 simulations per hour on premises, to running thousands of simulations per hour in the cloud.
UberCloud’s engineering simulation platform for manufacturing CAE applications, running in Microsoft Azure with the latest 3rd Gen AMD EPYC processors, is key for gaining a competitive advantage. You can bring better, more innovative products to market faster, shorten product development time, and increase your engineers’ productivity and agility, while lowering design and development costs.
Learn more about how UberCloud can enable your organization to take advantage of a multitude of available cloud resources for manufacturing.
This content was produced by UberCloud, Microsoft Azure, and AMD. It was not written by MIT Technology Review’s editorial staff.