John Gruber, creator of the Apple-centric media site Daring Fireball, recently pointed out that, in a recent Geekbench 4 processor test, the Apple (NASDAQ:AAPL) A10 Fusion chip inside the iPhone 7 Plus actually delivers better single-core processing performance than any Intel (NASDAQ:INTC) processor inside the MacBook Air.
Granted, the MacBook Air relies on processors from Intel that are now two generations old (they are fifth-generation Core processors; Intel recently released its seventh-generation Core processors), but the fact remains that even compared to Intel's best, the A10 is within striking distance.
If Apple is able to deliver another quantum leap with its A11 processor in next year's iPhone, then there is a real chance that -- for low-power applications, anyway -- Apple may actually deliver better per-core performance than even Intel's best low-power notebook chips.
The people at Intel aren't oblivious to this risk, though, and I am sure that they are going to try their very best to make sure that the chipmaker's future low-power processors will be able to stay ahead of the performance that Apple delivers with its A-series chips.
In this column, I'd like to go over the key strategies that Intel can pursue to stay a step ahead of Apple.
Faster dual-core chips for ultrathin devices
For very thermally constrained devices (such as Apple's thin and light notebook), it's unlikely that Intel will be able to simply increase core counts. The thermal constraints in such devices are quite similar to those of large phones and tablets, so sticking a third or even fourth high-performance processor core in a chip intended for this market just isn't going to fly.
Instead, Intel will need to aggressively focus on improvements in two key areas. The first is optimizing its processor cores so that they can run at significantly higher speeds in mobile power envelopes. Intel's top 4.5-watt sixth-generation Core chip (used in Apple's 12-inch MacBooks) can run a single processor core at up to 3.1 GHz. Its seventh-generation Core successor (which Apple has yet to adopt) uses the same basic architecture but can run a single core at a much higher 3.6 GHz.
Intel's next-generation ultralow-power processor will likely need to deliver both an improvement in the amount of work that the chip can do per clock and a further jump in clock speeds. Improving performance per clock by, say, 5% and bumping the maximum frequency of a single core to 4 GHz (an 11% improvement) should be able to do the trick.
More cores for more powerful devices
Although the "more cores" trick won't work in a processor targeted at a 4- to 6-watt thermal envelope, such a strategy may be viable for higher-performance chips intended for systems that can handle a more power-hungry chip.
Indeed, according to a recent leak, Intel is planning to introduce 15-watt and 28-watt processors suitable for devices such as the MacBook Air and the 13-inch MacBook Pro, respectively, with not just two processor cores, but four. Additionally, those chips will include Intel's Iris graphics.
This means that even if Apple manages to achieve single-core performance similar to Intel's low-power notebook chips, those processors should use the added thermal headroom to simply pack more cores in.
Intel can't regularly increase core counts, because at some point the power budget simply won't allow it (without serious sacrifices to per-core performance), but it will be very hard for Apple or any smartphone manufacturer to put that kind of multicore performance into smartphones anytime soon.
Interestingly, according to that same leak, Intel intends to move its products from four cores to six cores for its highest-power (and highest-performance) notebook processors (think 15-inch MacBook Pro class system). Again, in this case, it would seem that Intel is trying to distance the performance of its notebook products from the top mobile chips (such as Apple's) by cramming in additional cores.
A reasonable strategy
At the end of the day, Intel needs to make sure that the processors that it produces for notebooks, whether they are very thin and light, or bulkier and very high-performance, can deliver meaningfully better performance than those used in the top smartphones and tablets.
Perhaps just as importantly, though, the company needs to work with key software vendors to make sure that there's a lot of software out there -- software that's useful to a large portion of personal-computer users -- that can deliver a real benefit from those additional cores.
Can the chipmaker ultimately pull this off? Only time will tell.