Affiliate Disclosure
If you buy through our links, we may get a commission. Read our ethics policy.

Apple proposes OpenCL as high-speed computing standard

 

Apple has signed on to an industry-wide alliance that will see many companies, including some of the Mac maker's processor and video card suppliers, work together to develop an open format for accelerating specialized computing.

Dubbed the Compute Working Group, the organization managed by Khronos Group has been formed this week and includes Intel as well as AMD and NVIDIA, both of whom compete against each other to supply video hardware for Macs in addition to other PC makers. ARM, Motorola, and Samsung are also among the involved manufacturers.

The group's focus will be to develop widespread standards for calculating "heterogeneous data" regardless of whether it's onboard a conventional central processor or on a new wave of video cards that can handle some specialized calculations in addition to their normal graphics duties. Like the OpenGL standard, any future standard would be publicly documented and royalty-free.

While the new development body has yet to settle on a standard, Apple is the first to propose a complete specification. Khronos highlights the Mac maker's new Open Computing Language (OpenCL) language as a first candidate; in its early form, the specification would "liberate" video chipsets from having to perform only visual tasks and would coordinate both graphics hardware and multi-core CPUs in improving overall system performance in the future.

Apple's format could lead to accelerating physics and image processing chores. However, the broader effort wouldn't be limited to full-size computers and could lead to faster handheld devices as well.

Cupertino, Calif.-based Apple is only just making its first tentative steps into heterogenous processing. The company first announced its intentions to develop OpenCL with the preview of Mac OS X 10.6 at this month's Worldwide Developers Conference but significantly lags efforts by AMD and NVIDIA in promoting standards themselves.

The two semiconductor designers this week unveiled a string of video cards and dedicated processors that all can greatly increase the speed of computing under certain conditions, such as scientific research or physics in games. NVIDIA's GeForce GTX series is the most conventional and processes both graphics and general work, while its Tesla line and AMD's comparable FireStream are meant solely to speed up certain computing jobs in workstations and servers.

To date, none of these devices support Mac OS X, although Apple is confirmed as being aware of NVIDIA's technology.