About six years ago engineering software teams noticed a shift in Moore’s Law. While computational growth was de-accelerating on new CPUs, it was accelerating faster than expected on GPUs. Way faster. The first simulation software engineers to test their solvers on GPUs got mind-boggling results. Some reported up to 1000x faster results by relying on the massive parallel computation available on the newest “graphics boards.”
For the first time, the phrases “simulation and analysis” and “real-time results” were being used in the same sentence. Today many of the essential solvers used in aerospace, automotive, and other high-end engineering fields are taking advantage of GPUs.
Progressive engineering teams are taking it a step further, by utilizing Cloud-based engineering services. Why Cloud? Because CAE in the Cloud combines the parallel computational wizardry of GPUs with virtually unlimited amounts of memory. Problems which seemed out of reach due to in-house IT limitations are suddenly not only possible but practical.
Many of these Cloud-based simulation services let the customer choose between CPU-only simulation and GPU-enhanced simulation. A typical speed boost when including GPUs is about a 50% reduction in time to solution. Multiply that by hundreds of outcomes, and you suddenly have more created more time in the product development schedule.
Rod Mach (a great engineering name) runs TotalCAE, an IT consulting service that specializes in engineering applications. “Not every model or application will benefit” from moving computation to GPU-enhanced Cloud platforms. “But it is definitely becoming more common in the implicit realm.” For example, Mach says Dassault Systemès Abaqus “needs a teraflop and several million degrees of freedom before it is worthwhile to pack up the problem for the GPU.” Those numbers are high, but not exceptional. But even single-workstation simulation runs can benefit from a GPU boost. “If you only have a few cores — eight or less — adding GPU is a turbo boost. It will make a difference on smaller jobs. Every hour counts.”
The Future is Cloudy
Where is this marriage of GPU and Cloud taking CAE? It seems not even the sky is the limit. A Ph.D. dissertation in aeronautic engineering published in 2020 points the way. Marta Camps Santasmasas at the University of Manchester successfully moved a CPU-bound problem to GPU and Cloud.
Santasmasas started with a thorny problem in studying wind flow in urban built environments. Until now, the well-known Navier-Stokes (NS) equations for computational fluid dynamics (CFD) has been the solver of choice for such studies. NS is traditionally CPU-bound; most researchers in the field can’t assemble a high performance cluster big enough to do a complete analysis.
Santasmasas reached into the CAE library and coupled a Lattice Boltzmann (LB) solver running on GPUs to the existing NS solver running on CPUs. As she reports in the dissertation’s abstract, “The resulting coupled Navier-Stokes Lattice Boltzmann (NSLB) solver combines the accuracy and computing speed of the GPU implementation of the LB model with the stability, low memory consumption and mesh flexibility of the NS solver. Moreover, the NSLB model exploits the widespread availability of CPU and GPU hardware on desktop and workstation computers.”
Unleashing Simulation Power
Adding Cloud and GPU to simulation allows engineering teams to match the right technology to the job; you don’t have to wait for a budget increase to bring new hardware in-house. This isn’t only for the very largest manufacturers; SMBs are finding ways to accelerate their time to product using Cloud-based and GPU-enhanced simulation.
Consider Globe Trailers, a Florida-based maker of lowboy trailers with a reputation for highest quality. The trailers are built in Florida, but the engineering team is global. “We are trying to double production every year, and that’s difficult,” said Jeff Walters, VP of engineering. “We needed a software platform that would support our progress instead of hindering it.”
Globe chose Dassault Systemès 3DEXPERIENCE platform, including the Single Source of Speed application portfolio. Now everyone on the team, no matter where they are located, share the same access to powerful engineering and the same data sets. “Working across offices, having the information on the cloud makes it much easier to collaborate, and also helps spark innovation,” noted a company engineer. “Engineers are always thinking over their problems. If I come up with a new idea or solution, I can be anywhere and get on the 3DEXPERIENCE platform and start sketching a model. I might be troubleshooting issues in the shop with my tablet or I might be at home and log in from the Cloud. It’s always accessible, with the data I need.”
To learn more about 3DEXPERIENCE on the Cloud check out our page, here.