Faster Time To Insights From Physics Simulations

Advertorial The transformative power of modern engineering is nothing short of extraordinary. From the high-stakes worlds of motorsports and competitive sailing, where the slightest adjustment can mean the difference between victory and defeat, to the creation of safer and more fuel-efficient vehicles, HPC is helping to set new benchmarks and drive better outcomes.

The complexity of the physics involved in engineering demands a staggering amount of simulation and modeling. With so many variables at play, it is simply not feasible to test every single prototype in the real world. HPC allows engineers to perform complex simulations and move from the conceptual stage to full design faster.

One key component of the engineer's toolkit is high-fidelity computational fluid dynamics (CFD). This powerful methodology predicts how objects will interact with fluids and gases, and is essential to aerodynamics and aerospace analysis, as well as weather simulations and industrial system designs.

In recent years, high-fidelity CFD has surpassed lower-fidelity methods like Reynolds-Averaged Navier–Stokes (RANS) simulations due to its ability to provide more accurate and reliable data. This allows engineers to create more precise and effective models, that can dramatically reduce development time, and improve outcomes across a wide range of industries.

However, these advanced simulations also demand significant computing power. CFD is more computationally intensive than previous approaches which means engineers need access to a large amount of compute resources that can scale to CFD demands. On-premises HPC is one option, but it can be complex and time-consuming to manage.

Increasingly, engineering teams are turning to the cloud to handle their compute-intensive workloads. The cloud offers flexible scale, productivity, and agility, allowing engineers to focus on their work rather than worrying about limited resources or lag times in the job queue.

Cloud HPC evolves to serve cutting-edge engineering

Cloud computing has progressed to highly tailored offerings that meet the needs of a variety of users. One group that demands particularly high levels of performance is product development teams using engineering software. These teams must overcome several hurdles in pursuit of excellence – including large problem sizes, complex workflows, massive amounts of data, and highly collaborative work environments.

To address these challenges, Amazon Web Services (AWS) has pioneered on-demand infrastructure with value-added customization. With Amazon Elastic Compute Cloud (Amazon EC2), organizations can rent virtual servers, or virtual server instances, to run applications and modeling without the need for their own physical IT infrastructure. This is a far cry from the past, when the computing and latency demands of HPC typically necessitated on-premises infrastructure.

Amazon EC2 is designed with a range of instances optimized for different configurations of CPU/GPU type, memory, storage, and networking resources. This enables organizations to tailor their computing power to match the specific requirements of each application or workload, without the need for costly overprovisioning.

One instance type that is particularly well-suited for running HPC workloads like CFD are Amazon EC2 Hpc6a instances. Purpose-built to run HPC workloads, EC2 Hpc6a instances are powered by two 48-core 3rd Gen AMD EPYC™ "Milan" processors, which are built on 7nm technology for increased efficiency. They also support 384GB of memory across the 96 cores, providing 4GB of memory per core. These attributes make Hpc6a instances ideal for workloads like CFD that require improved network throughput and packet rate performance.

Product development is a rapidly evolving, highly competitive field in which the ability to process more data in less time confers significant advantages – both for engineers and end users who rely on the latest innovations to triumph on the track and beyond. But now, thanks to Hpc6a instances, product development teams can leverage the power of advanced silicon architecture and ultra low-latency inter-node bus technology to accelerate time to insights – and by doing so can reduce overall engineering costs.

At the heart of Hpc6a instance's performance lies AMD's EPYC processor, which is designed with a system-on-chip architecture. This means there's no need for additional chipsets on the motherboard, which helps to streamline processing. Additionally, AWS' Elastic Fabric Adapter (EFA) and its inter-node connectivity technology enables tightly-coupled workloads, such as CFD, with low latency to scale faster, ensuring that data-intensive processes run smoothly and efficiently.

Hpc6a instances in action

Emirates Team New Zealand is one example demonstrating the incredible potential of Amazon EC2 Hpc6a instances, with 3 Gen AMD EPYC CPUs inside, on the world stage. The competitive sailing team partnered with Ansys, to simulate their yacht's aerodynamics and hydrodynamics to assist in securing victory in the 2021 America's Cup.

Thanks to HPC in the cloud with Amazon EC2 Hpc6a instances, complex physics simulations like CFD are now accessible and affordable for product engineering teams around the world.

Sponsored by AWS & AMD.