Without advanced computer simulation, modern manufacturing as we know it today would be impossible. The complexity of today’s product designs and intense competition within and between markets are challenging manufacturers to be more efficient and more innovative than ever. If companies want to keep up, implementing a simulation-driven design process is the answer – because today, it’s usually the case that whatever manufacturer has the most capable high-performance computing (HPC) infrastructure, that’s who will be most successful.
Altair and AMD recently teamed up to put AMD’s latest 3rd Gen AMD EPYC™ 7003 series processors with AMD 3D V-Cache™ technology to the test by running a variety of real-world dynamic finite element structural and computational fluid dynamics (CFD) simulations.
Simulation is used in everything: From physical design, to thermal and fluid flow, to electromagnetic interference, and much more. Faster, higher-fidelity simulations produce better products, reduce defect rates, minimize manufacturing and warranty costs, and hasten time to market – all of which drive revenue and profitability. To maximize competitiveness, engineers need:
- Higher simulation throughput with fast turnaround times
- The capacity to run more scenarios involving higher fidelity models
- Lower infrastructure costs – including acquisition, management, and data center costs
- A secure environment to safeguard IP and protect systems from internal and external threats
Recently, Altair ran a series of multiphysics benchmarks with several models, including a realistic belt-sander model to demonstrate the effectiveness of Altair solvers and the latest AMD EPYC™ Gen 3 processors in meeting these challenges.
About the Benchmarks
The belt sander model used in the performance tests is comprised of multiple materials and approximately 280,000 elements. The simulations involved a “drop test” (dropping the sander from a height of 1.5 meters to a hard surface) and a motor cooling simulation that used CFD techniques to model heat transfer from the motor to the appliance shell.
Multiple industries utilize drop testing to develop more robust and durable products, including electronic devices, mobile phones, and household appliances. Manufacturers typically run thousands of these tests, virtually dropping devices from different heights and orientations to optimize material selection and physical design. The explicit implementation of the finite element method (FEM) in Altair® Radioss® is ideal for these simulations because it provides highly accurate results and minimizes simulation runtime – because of this, Radioss has become a go-to tool for manufacturers. It predicts dynamic and transient-loading effects on various structures to improve safety, survivability, and durability. And because design-phase simulation helps manufacturers avoid physical prototyping, they can reduce costs, and boost product quality without having to spend time building and testing real-world models.
A thermal and flow analysis benchmark was also conducted on the same model using Altair® AcuSolve®. This general purpose CFD solver is a massively parallel finite element Galerkin Least square-based code that provides stable simulations and accurate results for a multitude of physics and complex flows. Flow and thermal analysis are essential for product safety and performance, and it ensures the selected materials selected can withstand anticipated temperature variances.
Putting EPYC™ with AMD 3D V-Cache™ to the Test
The benchmarks were run on a dual-socket server with two 3rd Gen AMD EPYC™ processors, each with 64 cores and AMD 3D V-Cache™ technology. Each processor featured a total of 768 MB of L3 cache. The results for each test were compared with a similarly configured dual-socket AMD EPYC™ 7003-based system with 256 MB of L3 cache per socket. As an additional point of comparison, both benchmarks were also run against an 8-core workstation used as a reference to run the same simulations.
The latest release of Altair Radioss was used for the drop test simulations. The benchmark results are summarized below:
- For impact analysis, performance gains using new 3rd Gen AMD EPYC™ processors with AMD 3D V-Cache™ technology ranged from 1.1x to 1.8x faster than earlier AMD EPYC™ processors without the additional cache.
- For thermal and CFD analysis, performance gains using new 3rd Gen AMD EPYC™ processors with AMD 3D V-Cache™ technology ranged from 1.1x to 1.5x faster than earlier AMD EPYC™ processors.
- When running Altair Radioss specifically, the drop test simulation ran 23x faster on a single-node running 3rd Gen AMD EPYC™ processors with AMD 3D V-Cache™ compared to an 8-core workstation used as a reference.
The AMD EPYC™ Advantage
First introduced in March 2021, AMD EPYC™ 7003 series processors are suited to technical computing workloads.[i] Built on 7nm technology, these processors bring together high-core counts, extreme memory bandwidth, large cache sizes, and massive I/O, which enables exceptional HPC workload performance. These capabilities make AMD EPYC™ the world’s highest performance x86 server processor for technical computing.[ii]
The introduction of AMD 3D V-Cache™ technology extends the performance leadership of EPYC™ 7003 series processors even further, providing 3x the amount of L3 cache. With up to 768 MB of L3 cache and 804 MB of total cache per processor, the latest AMD EPYC™ processors offer exceptional performance for HPC workloads.
In addition to delivering industry-leading performance, EPYC™ processors also deliver advanced security features to safeguard customer data from internal and external threats. AMD Infinity Guard is a suite of advanced, multi-layer security features built into the AMD EPYC™ processor’s silicon that protects critical systems as they are booted, and as software is executed. These features make AMD EPYC™ systems highly resistant against today’s sophisticated cyberattacks, helping protect sensitive data, avoid downtime, and reduce resource drain.
AMD EPYC™ systems also offer exceptional power efficiency and density, delivering high levels of performance per watt. They help lower operating costs in HPC data center environments by minimizing server footprint and reducing power and cooling costs.
Compelling Business Results
With faster simulations, manufacturers can explore the design space more thoroughly using techniques like designs of experiment (DOE) and machine learning. These methods maximize durability and improve product safety while minimizing design and manufacturing costs.
Altair solvers coupled with 3rd Gen AMD EPYC™ processors and AMD 3D V-Cache™ technology address an industry “sweet spot,” delivering unparalleled performance for the medium-sized models typical in applications ranging from power tools to home appliances. Manufacturers can expect to see similar results across multiple industries and use cases, including crash and safety testing, automotive subassemblies, and battery pack development for e-mobility applications.
Using the latest AMD EPYC™ processors with Altair simulation tools, customers can compress the time required for engineering simulations without sacrificing accuracy or model detail. They can also run more thorough, more detailed analysis jobs in the same time, exploring the design space to produce higher-quality products while reducing manufacturing costs.
To learn more about AMD EPYC™ processors, visit https://www.amd.com/epyc.
[i] “Technical Computing” or “Technical Computing Workloads” as defined by AMD can include: electronic design automation, computational fluid dynamics, finite element analysis, seismic tomography, weather forecasting, quantum mechanics, climate research, molecular modeling, or similar workloads. GD-204
[ii] MLN-016: Results as of 01/28/2021 using SPECrate®2017_int_base. The AMD EPYC 7763 measured estimated score of 798 is higher than the current highest 2P server with an AMD EPYC 7H12 and a score of 717, https://spec.org/cpu2017/results/res2020q2/cpu2017-20200525-22554.pdf. OEM published score(s) for EPYC may vary. SPEC®, SPECrate® and SPEC CPU® are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.