Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Axemantitan

macrumors 6502a
Original poster
Mar 16, 2008
536
96
AMD's solution was simple: they brought PCIe outside the case, pin for pin. Sunnyvale contracted with JAE for a connector which could handle PCIe bandwidth at low cost, while remaining small enough to fit on thin laptop cases. The cable will run a short distance from a connector on the laptop to a similar connector on a breakout box, where the exact same circuitry as an internal card can perform the exact same function. In theory, there should be no performance difference between XGP systems and the same GPU in a desktop graphics card implementation. The XGP connector allows up to eight lanes of PCIe 2.0, and a larger x16 connector is in the planning stages.

This physical connector is dead easy for any laptop vendor to build into their laptops, and dead easy to design external devices for, as the circuitry is the same as PCIe. Nothing fundamental prevents it from being used on Intel or VIA laptops, or with NVIDIA GPUs, or even other PCIe devices. New drivers shouldn't even be needed. AMD, however, has done more than enable these devices. When the platform was announced in June, AMD also announced new drivers for their Radeon GPUs that would allow them to drive the laptop's internal screen with the external GPU on systems designed for such a feat. Further, they've provided reference designs for a few possible uses of the new platform, including single and dual GPU external graphics cards.

Could Apple seek a similar solution for future laptops?

http://arstechnica.com/news.ars/post/20081205-amds-xgh-external-gpus-may-soon-see-the-light-of-day.html

XGP_mockup.jpg
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.