In November 2018, Amazon Web Services (AWS) announed Elastic Fabric Adapter, a network interface for HPC applications running on EC2. We previewed the EFA technology, running benchmark CFD simulations with OpenFOAM, presented previously: (1) strong scaling – 97 m total cells, external aerodynamics around a car; (2) weak scaling – 100 k cells per core, flow over a weir. At 1008 cores, EFA delivered linear scaling in the strong scaling test and ~70% scaling in the weak scaling test. With linear strong scaling, EFA enables faster solutions for the same cost simply by running on more cores. The combination of open source software, C5n instances and EFA provides impressive HPC, at a price accessible to all.
CFD Direct From the Cloud™ (CFDDFC) is a Marketplace Product for AWS EC2, that provides a complete platform with OpenFOAM and supporting software running on Ubuntu Linux. It enables simple, fast creation of a cluster of instances to run CFD with OpenFOAM. We tested Amazon EC2 C5n instances, launched in November 2018, on 2 cases of steady-state, external aerodynamics around a car and transient simulation of water flowing over a weir. With meshes of 50-100 million cells, results showed 70%-90% scaling at 504 cores. The cost of the simulations was in the order of $100, compared to $100,000+ to purchase on-premises hardware and licences of proprietary software.
CFD Direct From the Cloud (CFDDFC) is our established platform for running OpenFOAM, ParaView and supporting software using cloud computing. CFDDFC is now available on AWS C5 and M5 instances of Elastic Cloud Compute (EC2). The CFDDFC product page on AWS MarketPlace now lists CFDDFC versions 5.2 and 4.2, corresponding to versions 5 and 4 of OpenFOAM, respectively. Users who launch CFDDFC with the command line interface should update their installation of CFDDFC immediately.
Cloud computing replaces large upfront expenses with low, variable payments that only apply to what you use. CFD simulations cover a range of sizes and complexity. Activity is generally interrupted by quiet periods when simulations are halted, e.g. when a simulation ends non-working hours, during analysis of results or preparation of a new simulation, etc. The fluctuating demand makes CFD well suited to a pay-per-use model. Users need to understand the costs involved to get the best value from cloud and be confident that they fall within budget. This article explains the costs, with an example of an external aerodynamics calculation with OpenFOAM using CFD Direct From the Cloud™ (CFDDFC) on Amazon Web Services (AWS).