Typically, whenever Apple announces a new iPhone, it also announces a new chipset. Beyond that, everyone inevitably compares Apple’s latest solution with offerings from Qualcomm, Samsung, Google, and MediaTek. It usually doesn’t take long for the benchmark results to come out and Apple to be declared the winner. But why do Apple processors always beat the competition? Why are the processors used by Android so far behind? Are Apple chips really that good? Let’s figure it out.

Does Apple release processors

Apple is developing processors that use the 64-bit Arm instruction architecture. This means that Apple’s chips use the same underlying RISC architecture as Qualcomm, Samsung, and Google. The difference is that Apple has an architectural license from Arm, which allows it to design its own chips from scratch. Apple’s first native 64-bit Arm processor was the Apple A7, which was used in the iPhone 5S. It had a 1.4GHz dual-core processor and a PowerVR G6430 quad-core GPU. At the same time, it was manufactured using a 28-nm process technology.

Apple’s latest mobile offerings use a six-core processor and their own GPU. The Axtal A15 contains 15 billion transistors, a 16-core Neural Engine and a video codec with support for HEVC and H.265 encoding and decoding, as well as support for MP4, VP8 and VP9 decoding. This processor is manufactured using TSMC’s second-generation 5nm process known as N5P.

In short, the latest generations of Apple processors offer better performance every time than any other smartphone processor from any company. On paper, scores for Apple processors (which only have 6 cores) are higher than scores for all eight-core processors.

Can you trust phone test results?

But here it is important to understand that Geekbench does not test those parts of the chipset that are responsible for the GPU, DSP, ISP and any functions related to AI. But they will affect the daily operation of any devices using these processors. But when it comes to pure processor speed, Apple is the clear winner.

iPhone scores very well in benchmarks

Here we need to dive into history a bit. It’s fair to say that Apple caught Qualcomm by surprise when it announced the 64-bit A7 back in 2013. Up to this point, Apple and Qualcomm have been supplying 32-bit Armv7 processors for use in mobile devices. Qualcomm led the way with its 32-bit Snapdragon 800 processor. It used its own Krait 400 core along with an Adreno 330 GPU. In those days, Qualcomm was doing well and relaxed a bit.

Is Apple better than Qualcomm?

When Apple unexpectedly announced a 64-bit Armv8 processor , Qualcomm had nothing. And the company’s management only had to call the 64-bit A7 a “marketing gimmick.” However, pretty quickly Qualcomm rolled out its own 64-bit system.

This happened in April 2014 with the release of the Snapdragon 810. A number of “Cortex” cores come directly from the custodians of the Arm architecture. But that same year, Apple announced the A8, its own second-generation 64-bit processor. It wasn’t until March 2015 that Qualcomm was able to announce its own first generation Snapdragon 820 64-bit processor with its own Kryo core.

In September of that year, Apple released the iPhone 6S with the A9 , a third-generation 64-bit internal processor. And just like that, Qualcomm suddenly found itself two generations behind Apple, although nothing foreshadowed this, it was just that Apple was developing very competently and did not allow them to be seriously leaked.

In 2016, Qualcomm again worked with solutions from Arm, but here everything was a little different. Arm has created a new licensing program that allows its most trusted partners to get early access to the latest processor designs and even customize them to some extent. The result was the Kryo 280 CPU core, of which there were eight in the Snapdragon 835, according to the data sheet. As for the launch of this processor, Qualcomm moved it from spring to winter. This means that the 835 was announced after the Apple A10 and iPhone 7.


Snapdragon – strength and power! But not in everything

This ping pong goes on. Things changed a bit when Arm introduced the Cortex-X line . These CPU cores were designed to close the gap between Android and Apple processors. Cortex-X is designed primarily for maximum performance, even at the risk of increased power consumption. This is why a mobile processor typically has only one Cortex-X core, then three high-performance Cortex-A cores, and then four power-efficient cores – a 1+3+4 arrangement.

How Apple processors are different from the rest

There are a few key things to know about Apple cores. She had the upper hand when it came to 64-bit Arm-based processors. Although Arm itself announced the release of the Cortex-A57 back in October 2012, Arm partners were expected to ship the first processors during 2014. Designing a high-performance mobile processor is a complex and time-consuming task. Therefore, it was not possible to pick up the novelty immediately.

Now the situation has changed, and the rhythm is such that Arm announces its new processor designs in late spring and this is enough for manufacturers to release their solutions in 6-8 months . Although, of course, they know about the plans to release new items about a year before us.

And Apple processors are expensive and large . This is also important in understanding their performance. The Snapdragon 888 is estimated to have about 10 billion transistors, while the Apple A14 has 11.8 billion transistors. A15 is even bigger – 15 billion transistors. The bottom line is that Apple sells smartphones, not chips. As a result, it can afford to make the chipset more expensive and recoup its costs elsewhere by slightly increasing the final retail price.

The characteristics of the A-series processors are really impressive.

However, Arm and Qualcomm are in the business of selling chips, and they need to make sure that the chips sell at a profit . And preferably with a big profit. This is the only way to make money on investments in research that are needed to release new products. Further down the road will be phone makers who also need to make a profit. So, you have to save on the processor at the very first stage, so that it is a compromise.

Apple processors have a large cache. Silicon costs money, and for some chip makers, their profits can be made for as little as 0.5mm2 of silicon saved. Here again, Apple can produce larger chips (in terms of silicon cost), including larger caches.

The Apple A14 has 4MB of L2 cache for each of the high performance cores and 16MB of system cache. The A15 uses an even larger cache: 6MB for each of the high performance cores and 32MB of system cache. These caches are huge compared to the Snapdragon 8 Gen 1 ‘s 2.5MB L2 cache for high performance cores or the 10MB L3 cache and system cache.

And Apple’s plan to build processors with wide pipelines at (initially) lower clock speeds has been a success. In a very broad sense, chipset manufacturers can either make a CPU core with a narrow channel but use that channel at high clock speeds, or use a wider one but at lower clock speeds. Just like a real water pipe, you can pump high pressure water through a narrower pipe, or lower pressure water through a wider pipe. In both cases, it is theoretically possible to achieve the same throughput. The processors created by Qualcomm follow the path of the “narrow pipe”, and Apple – the path of the “wide”.

Will Qualcomm be better than Apple?

One way Qualcomm could catch up with Apple would be to hire some former Cupertino engineers who worked on processors and have them develop Qualcomm’s processor. This is exactly what Qualcomm did.

Nuvia was a company founded in 2019 by former Apple head of processor design Gerard Williams and John Bruno, a systems architect at Google. They worked for Apple for five years in a similar position. Williams was the chief architect of the CPU at Cupertino . He has worked on the Cyclone, Typhoon, Twister, Hurricane, Monsoon, and Vortex CPU architectures for various Apple A-series chipsets. Before joining Cupertino, Williams spent 12 years as an Arm Fellow working on the Cortex-A8 and Cortex-A15 architectures. Qualcomm bought their company for $1.4 billion.

Since then, the former Nuvia team has been working on a new processor for Qualcomm. It will be its own design and its initial versions will be aimed at laptops. Qualcomm plans to release the first consumer products with Nuvia solutions in 2023. After that, Qualcomm will probably try to create a version for smartphones based on the same technology .

It’s undeniable that Apple has a world-class processor design team that has consistently produced the best chipsets in the world over the past few years. Apple’s success is not magic . It’s the result of great engineering, fast performance compared to the competition, and the luxury of building products that spare no silicon.


In any case, do not forget that each processor is the pinnacle of technology.

We can only get the best processor from Qualcomm, Samsung or Mediatek in a few cases. For example, if Apple for some reason releases a bad processor. So they will lose leadership, but only because of their mistake. One of the vendors mentioned might do a good job themselves and come out with an expensive processor with a lot of surface area and lots of dies for things like cache. But in this case, it will be very expensive, and smartphones will rise in price even more.

Are Apple processors really better than others?

Here, let’s go back to the beginning. I said that Geekbench does not test GPU, DSP, ISP and so on. Namely, these components in Apple processors are not so good compared to Qualcomm solutions.

Ultimately, it all comes down to user experience. Does the iPhone offer a good user interface? Yes. Does the latest Android flagship using the latest Snapdragon processor provide a good user experience? Also yes.

But all modern Apple, Qualcomm and Samsung chipsets contain specialized neural chips (NPUs). They perform tasks such as object detection, selection, recognition, face detection and identification. They do it much faster than the CPU, which is very good at Apple. The use of machine learning becomes a fundamental part of the user experience and does not depend too much on the power of the processor. It is on this field that the battles of manufacturers will unfold. Therefore, do not blindly follow the numbers in the tests. User experience is much more important, and here everyone needs their own.