Nvidia to expand CUDA parallel processing to mobile devices

CUDA could provide a big jump in the ability of smartphones and tablets to run multimedia-rich applications

Nvidia is looking to expand its CUDA parallel-processing architecture to mobile devices in the next few years, the company's CEO said Wednesday.

The move, when it comes, could mean a big jump in the ability of mobile devices like smartphones and tablet computers to run multimedia-rich applications.

[ InfoWorld tapped CUDA as one of 7 programming languages on the rise. | Get the best iPhone and iPad apps for pros with our business iPhone and iPad apps finder. | Keep up on mobile developments with InfoWorld's Mobilize newsletter and Technology: Apple newsletter. ]

"The benefit of CUDA in mobile devices is shocking," said Nvidia CEO Jen-Hsun Huang, in a Tokyo interview.

CUDA allows for certain processor-intensive tasks such as video editing or image manipulation to be off-loaded from the main processor to a graphics processor, realizing a significant improvement in data processing speed.

Graphics chips are restricted in the type of processing they can carry out but, with over 100 cores in a modern graphics processor versus just a handful of cores in a CPU, they are particularly good at the kind of image and video tasks that tax a typical CPU. The CUDA architecture combines software and hardware to realize these improvements.

CUDA was quickly adopted for supercomputers, and some of the fastest machines in the world are based on combinations of Nvidia GPUs and Intel CPUs. More recently it's arrived on the enterprise and consumer desktop and helped speed up tasks like video encoding in Adobe Premiere CS5.

"Imagine we take that CUDA capability and we put it into our cell phone," said Huang. He suggested an app that could recognize a bottle of wine from a phone's camera, and then interface with a cloud-based service to provide information about the wine, its best price and when it should be drunk.

"You're looking through this mobile camera and information about your world is popping up," he said. (See Huang's comments about CUDA on mobile in a YouTube video.)

Huang said his guess for when this will be possible is two or three years from now.

Nvidia's main focus in the mobile space is currently its Arm-based Tegra processor.

The dual-core Tegra 2 chip has already been used in some products, like Toshiba's recently launched Folio tablet computer, and is tipped for use in others including an upcoming tablet from Dell.

"The Tegra business is about to go through some very exciting times," said Huang. "The coming out party for Tegra is likely the Consumer Electronics Show in Las Vegas in the first week of January."

Huang said they'll be announcements of Tegra-based products at the show and then throughout the period until the Mobile World Congress exhibition in Spain in the middle of February. (See Huang's comments about Tegra at CES in a YouTube video.)

Martyn Williams covers Japan and general technology breaking news for The IDG News Service. Follow Martyn on Twitter at @martyn_williams. Martyn's email address is martyn_williams@idg.com,

Join the discussion
Be the first to comment on this article. Our Commenting Policies