DE Roundtable: What Would You Do If You Have On-Demand Access to a Supercomputer?

Alan Chalker from Ohio Supercomputer Center and Lee Margetts from simulation industry association NAFEMS join DE's editor Kenneth Wong for the roundtable talk on supercomputing.

What would you do if you have on-demand access to a supercomputer? How would you work or design differently? How would you approach your engineering simulation problems?

Previously, the cost of ownership made high-performance computing (HPC) or supercomputing the exclusive domain of well-funded universities, large enterprises, and government entities. But the emergence of on-demand HPC vendors, Open Source software for managing HPC clusters, and creative deployment of HPC-powered simulation apps may be changing the landscape.

Engineers developed workarounds -- like simplifying their geometry, scheduling long-lasting simulation jobs to occur during after hours, and reducing the number of alternatives in the design of experiment (DOE) studies -- because computing resource is limited. Over time, these practices have become so entrenched in the simulation industry that users follow them as standard practices, forgetting the reason behind them.

But now that HPC is becoming more affordable and accessible, long-held simulation practices should be reexamined.

In the upcoming DE Roundtable talk on April 12, 2PM Eastern (11 AM Pacific), Alan Chalker from Ohio Supercomputer Center and Lee Margetts from simulation industry association NAFEMS join DE's editor Kenneth Wong to discuss:

• Supercomputers that you can “borrow” or “rent” for the duration of a project;
• Open Source software to manage HPC
• Tackling large-scale simulation using in-house or on-demand HPC
• Deploying HPC-powered simulation as on-demand apps