A Cray supercomputer has set a speed record in a crash/safety simulation used by the automotive and aerospace industries. Leveraging the scalability of the Aries interconnect in the Cray® XC30™ supercomputer, engineers ran the “car2car” model, a 2.4-million element crash/safety simulation, in under 1,000 seconds. The LS-DYNA results are posted on topcrunch.org, the site that was created to track the performance of HPC systems on actual engineering applications codes.
In addition to the job turnaround record (931 seconds), the result set a record for the level of parallelism (3,000 cores). As the automotive and aerospace industries continue to run larger and more complex simulations, the performance and scalability of the applications must keep pace. Running the car2car model in under 1,000 seconds is a significant milestone in the ongoing effort to enhance performance.
During the past 25 years the model size for production crash/safety simulations has increased by a factor of 500. Initially simulations focused on single load cases (frontal crashes, for example), but today about 30 load cases (including frontal, side, rear and offset impacts) are investigated at one time. At the same time, time crash/safety simulation has progressed from being a field of research to becoming an integral part of a CAE Driven design process, enabled by the enormous increase in HPC compute power used by the auto industry. In recent years, the transition from SMP parallel simulations to MPI parallel simulations and the increase in processor frequency has provided the required performance. However, microprocessor speeds have now leveled off — so the increase in scalability, as demonstrated in the Cray car2car benchmark performance, is more important than ever.
The requirement for both application scaling (capability computing) and system throughput (capacity computing) continues to grow. The “THUMS” human body model has 1.8 million elements, and safety simulations of over 50 million elements are on the roadmap. Models of this size will require scaling to thousands of cores just to maintain the current turnaround time. The introduction of new materials, including aluminum, composites and plastics, means more simulations are required to explore the design space and account for variability in material properties. Using average material properties can predict an adequate design, but an unfortunate combination of material variability can result in a failed certification test. Hence the increased requirement for stochastic simulation methods to ensure robust design. This in turn will require dozens of separate runs for a given design and a significant increase in compute capacity — but that’s a small cost compared to the impact of reworking the design of a new vehicle.
Leading-edge automotive HPC environments already have petaflop-size compute capacity, but even that won’t be enough to meet crash/safety simulation requirements in the near future.