If you have been following our Blog you have seen our postings about how JARVICE helps enable our customers to build and deploy their own custom application on our HPC cluster and run them On Demand. You have also seen how easy it is to leverage JARVICE to spin up powerful 3D Workstations in the Cloud as well. What I would like to discuss today is how Nimbix has leveraged these capabilities and partnered with ANSYS to allow their users to leverage our HPC cluster to complete their work faster and run larger more accurate simulations.
ANSYS Portal on the Nimbix Cloud
Once a user signs up for an account they are able to purchase or transfer a license(s) from ANSYS or their current ANSYS partner and upload them to the cluster.
After the license has been added to the portal, the user can either upload their existing models or launch an interactive 3D desktop to create one from scratch and perform any necessary prep/post work.
Then users are able to solve their simulations across several (As many as their license supports) 56GB FDR Infiniband interconnected multi-core servers simply by specifying their input files, how many 16-core compute nodes they wish to use and where they want their results to be returned.
ANSYS Interactive Cluster
They also have the option of spinning up a dedicated ANSYS Interactive Cluster. The ANSYS Cluster provisions a secure, pay-per-use 3D graphics-enabled desktop and computation cluster for supported applications. Users are able to use the Interactive Cluster for pre/post processing as well as simulation and solvers using RSM.
So why is this important? It provides end users accesses to affordable HPC computational resources without having to go through all of the expense and time to build out an HPC cluster.
So why is it important to have access to affordable HPC resources? The problems these products are used to solve are growing in size and complexity while the time pressure to get answers is shrinking. I have spoken with several customers that explained that they had to “Simplify” their simulations in order to get their answers back in time with the equipment they have available.
Instead Nimbix enables end users to get access to true HPC resources with bare metal performance when they need them and only pay for what they use. It allows users to run those more complex simulations and still get the answers they need in the time available.
Take for example ANSYS Fluent. It scales linearly which means if a customer has a simulation that takes 8 hours (480 minutes) with 16 cores (1 node) you can scale up to 128 cores (8 nodes) and complete it in an hour (60 minutes). The infrastructure cost would be the same $14 to run, but you get your answers back 8 times faster.
The best part is because Nimbix bills to the minute if you scaled to 256 Cores you could get your answer in 30 minutes and pay the same $14 for the infrastructure!
The reason a company run simulations in the first place is to avoid costly production mistakes. If they are being forced to “Simplify” their simulations it increases the chance that a costly production mistake will happen. It also means they may be settling for a solution that works but may not be the best solution. If they are able to get results back quicker they can run more scenarios/options and find the optimal one rather than one that works.
If you are an ANSYS customer and currently limited by outdated/underpowered hardware capacity, tired of spending thousands of dollars just to maintain your systems, would like to work with larger models or be able to complete your simulations faster please contact us.