Curious about high-performance computing? You're in the right place!
Our FAQ page is designed to be your go-to resource for unraveling the intricacies of our cutting-edge HPC solutions. Whether you're a tech enthusiast, a seasoned professional, or someone just diving into the world of high-performance computing, we've compiled a list of frequently asked questions to guide you.
Frequently Asked Questions
Have Question? We are here to help
Why do we need high-performance computing resources?
Engineering and scientific applications like weather and climate, drug design, earthquakes, automotive and aerospace, and many other complex physics-based applications, are using highly complex software (CAE = Computer Aided Engineering) that does not fit anymore into any laptop or powerful engineering workstation, or they are running for many hours and even days. Only supercomputers and high-performance computing (HPC) resources in the cloud are able to handle these simulations in a reasonable time.
Why do we need cloud for engineering simulations?
In recent years, product design and development became more and more complex, with increasing product requirements and application complexities, such as digital twins, multi-physics, machine and deep learning, predictive maintenance, IoT, Edge, and much finer geometry decompositions with many millions of finite elements. Problems of this size do not fit anymore in high-performance workstations and midsize compute clusters, and the only solution is to move these applications to much more powerful compute resources in the cloud.
What’s the major difference between cloud and on-premise resources?
The major differences between cloud and on-prem resources result from the access to more, on-demand, and latest-technology HPC compute nodes, with the following major benefits:
- Shortening of the product development and thus time to market
- Improvement of product quality through detailed and faster parameter studies
- Thus, increase ROI and discover design failures early
- And strengthening the competitiveness and ability to innovate fast
- With cost savings by avoiding expensive on-prem hardware and maintenance
- On-prem computing resources are aging fast and get outdated quickly
- Integrability of Cloud HPC into the company IT environment (as part of its digital transformation) and thus abolition of internal IT silos
- And greatest flexibility with HPC as a Service and selection of any resources, increase in engineering productivity, and no learning needed.
UberCloud is a cloud services provider which offers software technology and services that enable engineers, at their fingertips, to access, through browser with simple login and password, and to build and run any simulation workload, on any cloud resources.
Is UberCloud a public cloud?
No. UberCloud is a cloud services provider offering software and services, on top of cloud resource infrastructure, that focus on easy migration of the engineer’s application workflow to any cloud, be it public, private or hybrid.
What are UberCloud Containers?
UberCloud containers are software that wraps engineering workflows into a secure high-performance virtual desktop enabling the engineer to perform geometry built, add physics, upload and download data sets, and submit simulation jobs to any type and size of compute cluster.
How secure is UberCloud (compared to public clouds and on-prem environments)?
UberCloud’s software layer adds two additional security components to the already highly secure infrastructures of the cloud infrastructure providers. First, on top, the UberCloud high-performance container which encapsulates securely the engineer’s simulation workflow and data; and the single-tenant cloud environment in the engineer’s own highly secure cloud account.
Does UberCloud’s offer include cloud hardware?
No. UberCloud is a cloud services provider helping engineers and their companies to move simulation workflows seamlessly to the cloud. UberCloud does this by boxing the engineer’s simulation workflow into a unique HPC software container, uploading it into the engineer’s own cloud repository, and moving it onto the compute cluster when needed.
How is UberCloud different from other HPC cloud offerings?
There are several major and important differences to other cloud services offerings. UberCloud is setting up a permanent private cloud in the engineer’s account on any hyperscaler’s cloud infrastructure, thus enabling sole and exclusive control of the engineer’s workflow(s) and data. Private cloud resources can be set up as needed and shut down when done, on demand, according to application and user requirements. And this private cloud can be set up in and moved to any public, private, or hybrid cloud environment, providing the same look and feel as the engineer’s familiar on-prem environment, no learning needed. That way, the engineer can benefit directly from the hyperscaler’s discounts, tools, and additional services, and from UberCloud’s additional security layers on top of the high security cloud providers offer.
How are the cloud offers of the Independent Software Vendors (ISVs) different from the UberCloud offer?
Large independent software providers (ISVs) tend to offer their own cloud services, limited to their own ISV software for their own customers. They tend to just offer latest software releases/versions on a limited number of compute instances (no GPUs), and simulation results have to be download to the user’s local machine, not stored in the cloud. Simulations tend to run in ISV’s own cloud without full control of the engineer not able to benefit from the cloud provider’s discounts. That makes ISV cloud offers ideally suited for small and medium size companies.
Does UberCloud charge for the engineer’s cloud resource consumption?
No. Instead, and unlike some of the other cloud services providers, UberCloud has a fixed software licensing model identical to the ISV licensing model, i.e., annual, and per seat. UberCloud’s annual software license cost is just a small fraction (like 5-10%) of the annual cost of an engineer consisting of the engineer’s labor, incidental, workplace, and IT cost. On the other hand, use of cloud resources can easily increase an engineer’s productivity by a factor of 10, i.e., 1000%.
How much does an engineering simulation job cost (on average) that’s running in the cloud?
A detailed Bill of Materials cost analysis for an automotive CFD simulation job based on 100 million cells, using STAR-CCM+ on a 16-node (1920 cores) Azure HBv3 cluster running 1.5 hours costs $122.46.
Is cloud HPC more expensive than on-prem HPC ?
It depends. If engineers use the cloud relatively excessively, it can happen that cloud resource usage exceeds the company’s cloud budget. Therefore, UberCloud provides tools and services for monitoring and reporting cloud usage per engineer, project, team, etc. In addition, cloud cost can further be reduced by committing to a certain amount of annual usage, and by using spot and reserved instances. That way, cloud spent is in the same range as on-prem total cost for HPC.
Is UberCloud replacing my company’s IT team?
No. UberCloud just builds a dynamic private cloud for the engineer and her/his company and provides maintenance and services to the IT team in all cloud related concerns and issues.
Have more questions or suggestions? We'd love to hear from you!
We strive to make your experience seamless and informative. If you have additional questions, feedback, or suggestions for improving our FAQ page, please don't hesitate to reach out. Your input is invaluable in helping us serve you better.