AMD banks on flood of Stream apps

GPGPU project expands to new markets

AMD is expanding its Stream project, which uses GPU cores to perform computing tasks normally sent to the CPU

AMD is looking to expand its Stream project, which uses graphics chip processing cores to perform computing tasks normally sent to the CPU, a process known as General Purpose computing on Graphics Processing Units (GPGPU).

Closely integrating GPU and CPU systems was one of the motivations for AMD's $5.4bn acquisition of ATI in 2006.

GPGPU systems have emerged in recent years as a favourite tactic for high-performance computing tasks.

By leveraging thousands of processing cores on a graphics card for general computing calculations, tasks such as scientific simulations or geographic modelling, which are traditionally the realm of supercomputers, can be performed on smaller, more affordable systems.

AMD Stream Computing director Patricia Harrell said that the company is now looking to bring the programme to the enterprise and consumer markets.

AMD will release a new driver for its Radeon series on 10 December which will extend Stream capabilities to the cards. Additionally, developers are being offered a software development kit which allows them to take advantage of the Stream system.

Harrell told vnunet.com that AMD hopes eventually to offer a complete and open platform on which software developers can easily optimise code for both the CPU and GPU.

"We envision a layer of industry standards that developers can choose from," she said. "This will be a really important step to encourage developers to take advantage of this capability and deploy software."

For everyday users, the advantages will range from dramatically improved video rendering times, to slight improvements in the performance of everyday productivity tools.

Terry Makedon, manager of software product management for AMD, told vnunet.com that even marginal improvements can add up for everyday office tasks such as reading large PDF files or viewing PowerPoint slides.

"You cannot really notice it, but by the end of the day you have saved 10 or 20 minutes of loading time," he said.

AMD also believes that the greatest benefits of Stream in the consumer and enterprise fields have yet to be uncovered. If developers latch on to the new system and optimise their code for Stream, Harrell expects new uses for the technology to emerge.

"It is my belief that there are killer apps we haven't even touched yet," she said. "There is a lot of possibility there."