Office applications Toolkit

Adobe takes to GridIron

Stephen Shankland CNET News

Published: 03 Feb 2004 09:55 GMT

Adobe Systems has signed a deal with GridIron Software that will enable its video effects software to tap into spare computing power on a network.

Under a licensing agreement announced on Monday, Adobe will include GridIron's XLR8 software with the next version of After Effects Professional, which is an Adobe package that lets editors spruce up videos with special effects, animated text and other modifications.

These effects often can require large doses of computing horsepower when applied to frame after frame of video. GridIron's XLR8 software can draw on unused processing power across the network to accomplish that task, GridIron said.

Software such as XLR8 is one step toward utility computing, which remains a highly sought goal of the information technology industry. In utility computing, a pool of hardware resources runs a constantly shifting collection of software, updated moment-by-moment to match a company's business needs. The hardware is used to its fullest capacity, instead of idling much of the time. However, the sophisticated management software needed is only beginning to emerge.

The "utility" moniker derives from the payment mechanism that could be used for such services -- a company could pay to use a third-party data centre the way it pays for electricity, water and other utilities. Similarly, a company division could pay for use of corporate resources.

GridIron's approach is often called distributed computing, which harnesses the power of PCs across a network. Distributed computing is the technology behind the SETI@home search for extraterrestrial radio communications and a successful search for large prime numbers.

Distributed computing and a close relative, grid computing, are becoming less rare. Pharmaceutical companies use it for drug discovery, chipmakers use it to create microprocessor designs and financial services companies use it to calculate the risks of stock portfolios.

Companies such as Platform Computing, Entropia and United Devices specialise in grid and distributed computing software, while computer powerhouses such as IBM, Hewlett-Packard and Sun Microsystems have their own similar technology. Wolfram Research's Mathematica calculation and visualisation software can harness grid computing, and Platform Computing has an extension that grid-enables Microsoft Excel.

GridIron got its start from technology developed to process large quantities of data produced by experiments from the Large Hadron Collider project at CERN, the European Laboratory for Particle Physics. The company released version 1.0 of its software in March 2003 and version 2.0 on Monday.

The first two versions work best with "embarrassingly parallel" computing jobs, in which a task can easily be split so it runs on a large number of computing nodes, said Gord Watts, vice president of marketing at GridIron. Version 3 is likely to be geared for more difficult "tightly coupled" tasks, which require much more communication between different nodes.

GridIron's software has been able to tackle some challenging distributed computing tasks, Watts said, including encoding video data into MPEG files.

XLR8 is designed to be embedded in other companies' software, Watts said. The Adobe deal "is the first of a number of customer announcements we'll be making in the next 60 to 90 days," he added.

Watts said the Adobe software would include permission to run the grid software on two processors of other computers besides the one After Effects Professional is running on. Customers will be able to purchase licences to run the grid software on additional processors as well, he said.

The grid component of XLR8 runs on machines with Microsoft Windows, Linux, Apple Computer's Mac OS X or Sun's Solaris operating systems.

Version 2.0 of XLR8 includes performance improvements and security features for ensuring that only authenticated applications can use a computer's resources, GridIron said. It also uses technology to rapidly share data across a network.