Credit: Hasan Kaymak
Naturally, Apple made its A13 chip even more powerful, leaving Android smartphones in the dust, and now if an analysis by Macworld’s Jason Cross is accurate, the A14 chip that’s expected to appear in this year’s iPhones is going to surpass even Apple’s highest-end 15-inch MacBook Pro, which is already a powerhouse.
Like everything else about its upcoming iPhones, Apple doesn’t say much about the chips that it plans to include, but as Cross notes, it takes many years to design chips and get production ramped up, so we’ve already heard leaks from suppliers like TSMC on some of the high-level specs that could be used for Apple’s A14 SoC.
5nm Manufacturing and What It Means
Taiwan Semiconductor Manufacturing Company, or TSMC, is the supplier that produces all of Apple’s A-series chips, and as Cross explains, Apple continues to benefit from improvements in TSMC’s manufacturing processes that allow for more power to be packed into smaller spaces.
The A12 and A13 chips were both built using a 7nm process, and while that was slightly enhanced for the A13 to provide faster speeds and better power efficiency, this year’s A14 is expected to make the jump to a brand-new 5nm process, and could very well be the first chip of its kind to use it.
This is a big upgrade. The 5nm mode is not a half-step by any stretch, but it is the next “full node” after 7nm. It uses extreme ultraviolet (EUV) lithography extensively throughout the process, and TSMC says it delivers 80 percent more logic density and can run either 15 percent faster at the same power as its 7nm chips, or 30 percent lower power at the same performance level.
Jason Cross, Macworld
These processes represent how small the transistor circuits can be made on a given CPU design. Since modern CPUs are made up of billions of transistors, the smaller they can be made, the more you can pack in. More transistors equals greater performance, and in addition, smaller transistors require less power.
The “nm” in this case stands for “nanometre” which is a billionth of a metre, or a millionth of a millimetre, so we’re talking about something that’s practically microscopic in size, and while it’s hard to believe that two millionths of a milimetre makes that much difference, if Apple keeps the A14 the same physical size as last year’s A13, this could allow it to almost double the number of transistors on the chip.
The A13 had 8.5 billion transistors, and according to Cross, if TSMC’s new specs are to be believed, the A14 could potentially include 15 billion transistors. That would put it above almost every high-end desktop CPU and GPU available. However Cross notes that this number is so crazy that he thinks it’s more likely that Apple would take advantage of the 5nm process to shrink the chip area a bit, which would still leave it with plenty of room to include well over 10 billion transistors.
Although it’s worth noting that the number of transistors isn’t by any means the sole measure of a CPU’s power — it’s how you use them that counts — it does provide an indication of what Apple will be working with, and would allow for things like more cores, bigger cores, and more cache. Basically, it gives Apple a larger surface to build on.
Cross notes that the single-threaded performance trend from recent A-series processors has increased at a fairly consistent rate, and with the A13 offering a 20 percent improvement over the A12, he predicts that the A14 will provide a similar speed boost.
The purported 15 percent clock speed boost afforded by the 5nm process alone would give us around 1,530. My guess is that Apple will likely wind up in the 1,800 range, due to both higher peak clock speeds and some architectural improvements made possible by the much higher transistor budget.
Jason Cross, Macworld
Multi-core performance is more difficult to predict with any accuracy, Cross says, but if Apple follows the trend line, the A14 in the iPhone 12 could easily hit a Geekbench 5 score of around 4,500, placing it well above the faster Android phones on the market, which only come in at 3,000, and if Apple pushes the process even a bit it shouldn’t have any difficulty reaching 5,000, which puts it in in the same territory as the 15-inch MacBook Pro and six-core mainstream desktop CPUs.
Cross also adds that he expects Apple will use the new process to push GPU performance even higher for gaming purposes, since Apple Arcade is making iPhone gaming more important than ever, and is estimating a possible 50 percent improvement in this area, thanks also in part to the rumoured 6 GB of RAM that the iPhone 12 models are expected to pack in.
The new process could allow Apple to add dedicated Neural Engine cores as well, which would push the machine learning performance to even higher levels — an area that’s become a huge priority for Apple’s recent iPhone models as they perform more and more computational photography tasks, along with Apple’s privacy-focused stance of performing analysis of user data such as photographic face detection directly on the user’s device rather than in the cloud.