Comparing the A14 Bionic to a 15-inch MacBook Pro's 6 core CPU is meaningless, and here's why
The work that Apple has put into the A Bionic series is, from the outset, exceptional. The A13 Bionic outperforms its competitors in almost every way and makes the iPhone 11 series some of the most powerful smartphones on the market. The A14 Bionic, likely to be found in the iPhone 12 or iPhone 11s series later this year, will undoubtedly offer better performance than the A13 Bionic and its predecessors. However, to argue that it could put a "15-inch MacBook Pro's 6-core CPU to shame" is a fallacy and overly simplistic.
The claim centres on the analysis of Jason Cross at MacWorld, who published a detailed report on "What we might expect from Apple's A14 Bionic processor". The article goes into depth on the effects that switching to a 5 nm node could have for Apple's next SoC, and is well worth a read if you are interested in understanding the differences between the 5 nm and 7 nm processes.
In summary, TSMC has achieved an 80% greater logic density with its 5 nm node, allowing company's like Apple to have approximately 15 billion transistors on a 100 mm² chip, a size that Cross settled on as being good for a "high-performance premium mobile processor" with numerous stacked components. By comparison, the A13 Bionic has 8.5 billion transistors spread across 98.5 mm².
Cross notes that transistor volume does not represent processor performance, but it is indicative of how much space Apple has for cores and cache, among other components. TSMC's 5 nm node is apparently 15% more powerful at the same wattage as its 7 nm predecessor, though. Additionally, 5 nm can achieve 7 nm performance levels but at 30% lower power consumption.
There has been no confirmation that the A14 Bionic will be based on a 5 nm process, but Cross has used it as a working assumption for analysis. The argument is convincing too, with TSMC due to start mass-producing 5 nm nodes from Q2 2020 onwards. The timeline of major flagship smartphone releases would probably mean that we will have to wait until the end of this year or Q1 2020 before we see Android OEMs using 5 nm SoCs, though.
In his analysis, Cross states optimistically that the A14 Bionic could score 5,000 in the Geekbench 5 Multi-Core benchmark. A score this high would eclipse most CPUs in our database, including the A12X Bionic found in the third-generation iPad Pro series. It would also outscore the Intel Core i7-8850H, the 45 W and 6-core CPU found in the MacBook Pro 15 2018. The Core i7-8850H averaged 4,860 in the same benchmark, for reference. Cross has been slightly misquoted here though, as he does note that predicting multi-core performance for the Bionic A14 is difficult. He points out that a score of 4,500 would be within the trendline of multi-core improvements from the A11 Bionic to the A13 Bionic too.
Regardless, the notion of using synthetic benchmarks to crown a new A Bionic series chip as being more powerful than its x86 counterparts is nothing new. In 2017, 9to5Mac proclaimed that the iPhone 8 was "faster than a Core i5 13-inch MacBook Pro" by referring to Geekbench analysis conducted by Tom's Guide. Comparing devices in this way may offer "insights as to the architectural comparison of the CPU in the system" as Forbes noted in 2015, but is not a good way to compare an iPhone against a MacBook Pro or any other x86 machine.
Simply, Geekbench does not measure a device's actual performance, actual being represented by real-world usage. Instead, it just tests "components of the SoC or some other part of the whole system". So, the A14 Bionic may compare to something like the Core i7-8850H in isolation, but that comparison quickly breaks down when one steps outside of a synthetic benchmark. The A14 Bionic, like all mobile SoCs, will be optimised for power consumption to a far greater extent than the Core i7-8850H will be, for example. In practice, this means that the former cannot maintain high clock speeds for as long as the latter can.
In short, benchmarks like Geekbench offer a snapshot of what SoCs can offer in an ideal scenario. They cannot offer a worthwhile comparison between architectures like ARM and x86, nor are they intended to do so. Implying that the A14 Bionic matches the performance a Core i7-8850H ignores any restrictions that are placed on the former, whether that be the other hardware with which it is paired or that is passively cooled.
Undoubtedly, mobile processors are closing the gap to their desktop and laptop counterparts. Hardware is only one part of the equation for ARM processors to offer comparable performance to x86 chips, though. As we have seen with Windows 10 on ARM and devices like the Surface Pro X, software optimisation plays a huge part in how powerful a device and its processor feel.