Nissan Drives 20 Percent Lower Costs on Crash Test Simulation with AMD EPYC™ Processors

Simulated vehicle crash testing allows Nissan to innovate faster and put safer vehicles out on the road — over three million globally every year, in fact. As the production process grows in complexity, Microsoft Azure VMs powered by 4th Gen AMD EPYC™ CPUs allow it to increase performance and keep computer aided engineering (CAE) costs manageable.

Continuous Improvement

Proliferating vehicle development requirements, from new electric drivetrains to more stringent safety regulations, mean Nissan’s engineers must run and re-run thousands of simulations long before a vehicle prototype gets put to the test on a physical proving ground. Enhancing the CAE capabilities of these users means continuously increasing HPC resources and software use. But with crash simulation testing the most expensive workload in Nissan’s CAE suite, and the software licensed based on usage time, that also risks escalating costs.

In 2021, Nissan switched from running its CAE simulations in its own on-premises data center to the cloud. This allowed it to increase and decrease HPC capacity in line with workload demands. This was a major benefit over its on-premises set up but, as you’d expect from one of the world’s leading automotive engineering teams, continuous optimization is always on the agenda.

So it wanted to find out more when the Microsoft Azure team said that HPC VMs with AMD EPYC processors could provide higher performance than its incumbent cloud provider.

Faster simulations, Lower costs

Nissan’s strenuous testing using benchmark data and CAE software simulations revealed that VM performance was 30 percent better with AMD EPYC CPU-powered Microsoft Azure instances compared to its previous cloud provider. The cost of each VM instance might have been higher, but the faster performance meant it could complete simulations in less time, driving 20 percent lower costs overall for HPC crash simulation. At Nissan’s scale, that adds up to cost savings in the millions. The 30 percent increase in simulation speed also allows the engineering team to quickly feed results back to product design, accelerating innovation.

The migration was completed in just six months, including developing the CAE environment on Azure and internal testing to evaluate crash simulation software performance and environment stability.

Just the Start

Nissan’s considerable Azure cloud fleet of HBv4 VMs powered by 4th Gen AMD EPYC CPUs has transformed the performance of its CAE software. The exceptional core density it now has access to — up to 176 cores per VM — means it has slashed its VM count and therefore its number of software licenses.

And that’s just the start. It’s now planning a full migration of its entire crash domain simulation to an Azure AMD environment.

Read the complete customer case study to learn more about how Nissan is driving a range of impactful meaningful business outcomes with AMD EPYC processors.