High-performance computing turns to apps to cut cost and frustration

A new model emerges for putting HPC into the hands of users

SALT LAKE CITY - Steve Jobs was right about apps in more ways than perhaps he ever knew. The concept of using apps to make software easily available and affordable to large numbers is arriving in highperformance computing (HPC).

There is a new effort to take code that contains a very small, self-contained calculation designed for one thing, such as modeling and simulating the flow of a fluid through a pipe, and turn it into an app.

The potential of apps in HPC is creating excitement in an area long use to frustration. Although there are ample examples of U.S manufacturers that have reduced time to market by using modeling and simulation, adoption of HPC tools faces formidable barriers for all the but the largest manufacturers.

HPC hardware and software can easily cost six figures, and even if a company makes the investment it may have trouble finding the expertise to run the systems.

But instead of buying large systems that can do many things, the equivalent of a Swiss Army knife, the app concept is focused on producing a piece of software that can do one thing.

"I think this can be the disruption to what's been a pretty stagnant software world for the last decade," said Tom Lange, Proctor & Gamble's director of modeling and simulation, of the potential of apps.

Lange believes that apps can help "democratize" HPC and make simulation analysis "an everyday part of decision-making."

P&G, a consumer products giant, uses HPC to make many of its products. It has the expertise, the money and connections to national laboratories to develop the codes it needs, but P&G is also making some of its code available as apps for this emerging effort.

This app model is already being used through NanoHub.org, a National Science Foundation supported platform. Users can get access to research software code to do things such as model nanoscale electronic devices. There are more than 250 codes, mostly available as "open access," but not open source, which would allow users to modify and change the code.

George Adams, the deputy director of NanoHub, and who is now director of ManufacturingHub.com, which will be source of these apps through the National Digital Engineering & Manufacturing Consortium (NDEMC) backed effort. Federal, state and private sector funding is supporting the initiative.

One problem addressed by the apps is the need for expertise. "The challenge with expertise in artificial intelligence-type software is there are potentially lots of decision points," said Adams. "If you focus your app on a narrower application area, the number of decision points is reduce and you do a better job of incorporating the expertise into the software."

Patrick Thibodeau covers cloud computing and enterprise applications, outsourcing, government IT policies, data centers and IT workforce issues for Computerworld. Follow Patrick on Twitter at @DCgov or subscribe to Patrick's RSS feed. His e-mail address is pthibodeau@computerworld.com.

See more by Patrick Thibodeau on Computerworld.com.

Read more about high performance computing in Computerworld's High Performance Computing Topic Center.

Copyright © 2012 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon