TGR-E, a leading engineering firm, recently migrated its on-premises HPC cluster to Amazon EC2 instances optimized for high-performance computing (HPC). By utilizing the HPC Optimized instance family, including Hpc6a instances powered by AMD EPYC processors, Hpc7a instances with 4th generation AMD EPYC processors, and Hpc6id instances powered by Intel Xeon Scalable processors, TGR-E gained the computational power required to efficiently run complex simulations at scale. These simulations encompass various computer-aided engineering tasks such as fluid dynamics, load analysis, and finite element modeling.
To facilitate the deployment of HPC architectures for application software, TGR-E leverages NICE DCV for remote desktops, creating a unified environment on AWS that combines both commercial and open-source engineering software. These simulations aid engineers in optimizing the performance of motorsports vehicles by analyzing the behavior of liquids and gases with surfaces, essential for improving aerodynamics and cooling systems.
The adoption of AWS Parallel Cluster simplified the management of HPC clusters for TGR-E engineers, automating resource provisioning and job scheduling to optimize performance. By utilizing Amazon EC2 instances with EFA for enhanced network connectivity, MPI jobs now benefit from reduced latency and improved network bandwidth, resulting in faster HPC job execution without additional costs.
For data storage and protection, TGR-E utilizes Amazon S3 for object storage and Amazon FSx for Luster for shared storage with the performance of the Luster file system. This setup allows engineers to write simulation results to Amazon S3 buckets seamlessly, enhancing data management and security.
Since transitioning to AWS, TGR-E has observed a 10 percent increase in processing speed from day one, alongside consistent and reliable performance with no downtime. The company has also achieved significant cost savings by eliminating traditional data center expenses such as rack space, cooling, and power consumption, resulting in more efficient workflows and operations overall.
Article Source
https://aws.amazon.com/solutions/case-studies/toyota-gazoo-racing-europe-case-study/