It's been nearly two years since Intel (INTC 2.20%) executives first officially mentioned the code name Broxton. Broxton is an upcoming mobile-focused system-on-chip from Intel intended to be competitive in high-end smartphones and tablets. The chip was first expected to go into production in mid-2015, but there's a lot of evidence floating around suggesting that the production -- and thus the launch -- date has been moved out to some point in 2016.
Interestingly enough, the first details of an Intel system-on-chip known as Apollo Lake-I, which should be very similar to the Broxton part that's aimed at phones and tablets, recently hit the Web. The details reveal something a bit unexpected.
Apollo Lake-I seems a bit light on the graphics cores
The predecessor to Broxton, known as Cherry Trail, delivered a solid boost in graphics performance relative to the older generation Bay Trail chip. A major part of that was due to a new graphics architecture coupled with a substantial increase in graphics cores -- there were four in Bay Trail and 16 in Cherry Trail. Despite this large boost in resources, Cherry Trail still delivered graphics performance well short of what the top mobile chips in the market today offer.
Interestingly enough, the Apollo Lake-I chip is expected to increase the number of graphics cores integrated onto the chip from 16 in Cherry Trail to 18, which I admit is a smaller increase than I might have expected. Now, there's far more to comparison than just number of graphics cores. Broxton should have a new GPU architecture and access to much greater memory bandwidth, which should help performance.
I'm just a bit surprised that Intel isn't going bigger in terms of graphics cores.
Pay attention to Gen. 9 GPU details at the Intel Developer Forum
At the Intel Developer Forum, which is expected to be hosted in August, the company is hosting a technical session dedicated to detailing the Gen. 9 GPU architecture. If the Gen. 9 graphics architecture delivers a substantial improvement over the Gen. 8 architecture, then "just" 18 graphics "cores" might be enough to be competitive in the high end of the market.
Why is this important?
One of the key areas of mobile system-on-chip performance today is graphics. And unfortunately for Intel, it's also one of the areas where Intel has been weak relative to the competition. Intel's own graphics architectures have been so behind the competition in mobile devices that even a full generation advantage in terms of manufacturing process technology relative to peers hasn't been able to enable leadership in high-end mobile graphics performance.
What's interesting, though, is that for Broxton -- Intel's high-end mobile system-on-chip for both phones and tablets -- the company is using its home-grown graphics architecture. In prior generations, Intel had two separate product lines for phones and tablets. The tablet chips -- Bay Trail, Cherry Trail -- have used Intel-designed graphics, while the phone chips have tended to use graphics from Imagination Technologies.
An optimist might look at Intel's use of Gen. 9 graphics in its high-end phone/tablet lines as an expression of confidence in its new graphics architecture. After all, Intel could very well have gone and licensed Imagination's latest PowerVR Series 7XT architecture and called it a day. A counterargument to that, though, is that Intel doesn't want to spend the development dollars working on drivers for another graphics architecture when it's already doing the driver work for the Gen. 9 GPU for PC processors anyway.
Once Intel launches the Skylake processors that feature this graphics architecture, we should have a pretty good idea of how it will fare in a low-power mobile processor. It shouldn't be too long now.
