Intel says its new processors with AMD graphics will be available in early 2018. But it looks like someone’s already testing some early samples, because a few photos of what looks like an Intel NUC mini PC motherboard with an Intel+AMD chip are making the rounds.
Among other things, this seems to confirm that the new chips are destined for compact desktop computers as well as thin-and-light laptops.
The upcoming chip or chips are expected to feature a custom GPU developed by AMD and an Intel Core H-series quad-core processor. The idea is to provide higher performance graphics than you’d get from a CPU with integrated Intel HD graphics, while taking up less space and using less power than a typical CPU + discrete graphics solution.
It remains to be seen just how competitive the new chip will be in terms of price or performance. This year AMD rival NVIDIA launched the Max Q platform which allows PC makers to build thin, light, and relatively low-power laptops with NVIDIA graphics.
But this could just be a first step for Intel. The company also recently hired AMD’s former GPU chief and announced that he’ll be heading up a new “Core and Visual Computing Group” that could lead to new GPUs for use in laptops, desktops, IoT devices, and other applications that can leverage GPUs for not only graphics, but also other tasks that rely on parallel processing such as machine learning, computer vision, or cryptocurrency mining, among other things.
Intel and amd on the same chip was one of the warnings of the apocalypse.
This would be nice for NUCs etc with weak GPU.
Oh, okay. How is this different from giving you two chips and two sockets on your motherboard? Do they have at least an internal bus between the chips on the die?
The CPU and the GPU modules are too close.
What is this, designed with 7 inch tablets in mind?
A typical Ultrabook/Convertible is 13 inches. That’s a lot of space to separate the two heat producers.
Which would mean much better passive heat dissipation, that means lower throttling, more ergonomic to the touch, and higher performance.
For once, I agree with Nvidia’s MaxQ philosophy… they used the tools they had in-hand, but managed to overcome most of the issues of performance in tight packages.