Apple laptops to feature Nvidia chipsets and two GPUs

Posted at: 7:08pm 14th October 2008 by Ben Hardwidge

Nvidia builds a chipset and GPU into one die for Apple’s new line of MacBooks, with a further GPU available for a boost

An Apple story in Custom PC – what’s the world coming to? Okay, so there isn’t really much that you can customise in a Mac, but Steve Jobs’ latest announcement could herald a major boost for GPGPU computing. The company has just announced that it’s ditching Intel chipsets for a new chip from Nvidia, and that its latest MacBooks can also feature two GPUs.

The new Nvidia chip combines a chipset and GPU in one die, with a GeForce 9400M taking up 70 per cent of the die area. It’s not a powerful GPU in 3D terms, offering just 16 stream processors, but Jobs claims that it offers 54GFLOPS of processing power, and says that it’s up to five times the speed of Intel integrated graphics.

The new chip will come with all of Apple’s new MacBooks as standard, including the Pro and lightweight Air models. However, the Pro models can also be upgraded to feature a GeForce 9600M with 32 stream processors, using 256MB or 512MB of GDDR3 memory. Seeing as this is a discrete chip on top of the existing chipset, this could give you 48 stream processors for parallel processing.

Okay, so this isn’t going to make a big impact in the 3D gaming world, but Apple is also going to be introducing its C-based GPGPU API, OpenCL, in the forthcoming Snow Leopard version of Mac OS X. With an Nvidia in all of its MacBooks as standard, with the option of adding more cores, it looks as though Apple has some big plans for GPGPU processing in its next OS. If Apple is taking it seriously, then it looks as though GPGPU is going to start to make a big impact in the future.

The new MacBooks will also feature an LED-backlit display, a mini DisplayPort connector and a glass touchpad that supports multi-touch gestures with up to four fingers. The Air will also come with a 128GB SSD. Could Apple’s use of two GPUs in its latest laptops herald a new era for GPGPU computing? Let us know your thoughts in the comments section below.