Over the past 30 years Computational Fluid Dynamics (CFD) has grown to become a key part of many engineering design processes. From aircraft design to modelling the blood flow in our bodies, the ability to understand the behaviour of fluids has enabled countless innovations and improved the time to market for many products. Whilst the accuracy of CFD remains a concern, the scale and availability of High Performance Computing (HPC) resources is arguably the main bottleneck for the typical CFD user. This need for greater HPC resources has accelerated over the past 5-10 years with the increasing desire to move to higher-fidelity approaches, as well as the growing use of optimization techniques and machine-learning driven workflows.
Over my career I can’t think of a single engineering group who didn’t complain about jobs being queued on their cluster. E-mails were frequently sent to the sysadmin to bump up their priority and arguments often started between rival engineering teams on who needed to run first! When I worked in Formula 1 the logic was simply that if you couldn’t get a job back by the end of the day then you missed the opportunity to analyze the job and get another queued by the end of the day. If you scale this up over the course of the year you could waste many days and potentially make the difference between winning or losing.
With HPC on AWS, capacity and queues should not be a bottleneck for engineers. You should be able to run your code to the limit of scalability rather than sizing your workload just to fit in the cluster queueing system. Major companies such as Formula 1 have been able to reduce their turn-around time from 60 hours to 12 hours simply because of the scale of AWS compared to their on-premises alternative.
Join me for a webinar on Computational Fluid Dynamics for Motorsport on 5th May to learn how new AWS compute services like Amazon EC2 C5n instances and network interfaces like Elastic Fabric Adapter (EFA) are transforming the performance of CFD codes on AWS.
Visit the Computational Fluid Dynamics on AWS page to get a $100 AWS credit to test out how your CFD code runs on AWS.