Regarding Apple‘s core-making team, we hear mostly praise. They have worked hard for ten years, and the debut of the M series is the pinnacle.
From a hidden and quiet corner to the turbulent spotlight, Apple’s core-making team under the leadership of Johny Srouji, Apple’s senior vice president of hardware, has also begun to be known to the world.
Since the release of M1 in 2020, Johny Srouji has also frequently appeared in public, telling the world about the excellence of Apple’s self-developed chips.
When we thought that Apple would take another step forward in the core-making milestone with the M chip, the A16 Bionic, which was unveiled in September this year, flashed through the Keynotes such as performance and specifications, without much ink.
Subsequently, with the launch of the iPhone 14 Pro series, in specific tests, the A16 Bionic squeezed a tube of toothpaste without success.
Even in the standard version of the iPhone 14, Apple has rarely used the “old chip” of the A15 Bionic, breaking the rules of a new machine with a new chip.
In this regard, there are many speculations from the outside world. Some estimate that Apple’s move is actually for cost considerations, and some say that TSMC’s foundry has insufficient production capacity.
A recent report basically answered the phenomenon of Apple’s new machine and old core, A16 Bionic squeezed toothpaste, etc. from the front, and even revealed some problems exposed within Apple’s core-making team.
A16 Bionic, rather than A15 Plus
As early as September before the birth of the A16 Bionic, ARM announced a new generation of public version architecture. The CPU was enhanced around optimization, while the GPU revolutionized the architecture and gradually introduced support for “ray tracing”.
Sure enough, in the flagship chips of the Android camp such as Qualcomm Snapdragon 8 Gen 2 and MediaTek Dimensity 9200 released at the end of the year, the high performance of the GPU and the popularization of “ray tracing” seem to have entered the “ray tracing era”.
More than that, with the previous generations of thunderstorms, the energy efficiency performance of the new generation of SoC has also reached a new level, which in turn raised the performance of Android flagships at the end of the year to a higher level.
Back before the release of the A16 Bionic, Apple’s core-making team was ambitious because they had introduced many innovative specifications, performance, and features on their GPUs, including support for “ray tracing”.
And in the process of development, A16 Bionic has considerable processing power. As long as it goes on the market as scheduled, I don’t know how much higher it will be than those “piercing the sky” functions.
However, in the prototype stage, due to engineering design flaws, the GPU of the A16 Bionic could not calm down. With the high-frequency call, the temperature also soared, and at this time it was accompanied by higher energy consumption.
In the end, the GPU of the new architecture was abandoned by Apple, and instead chose a safe upgrade strategy, which was taken from the A15 Bionic GPU, made a simple iterative upgrade, and finally became the SoC on the iPhone 14 Pro series.
The new GPU that was originally expected to appear on the iPhone 15 (that is, the A17 Bionic) is not ruled out, provided that the Apple core-making team finds out how to fix a series of problems such as power consumption and heat accumulation under high load.
The change of design also made the A16 Bionic, which would have drawn a lot of color in the history of A-series chips, become a dispensable existence.
Such a change is also called an “unprecedented” failure in the history of Apple’s core manufacturing.
This also affects the formation of the new iPhone from the side. In order to deliberately widen the gap between Pro and non-Pro, Apple is also “forced” to equip the iPhone basic series with the old A15 Bionic chip.
Hidden worries behind Apple’s core-making team
Replacing the GPU design of the A16, like a butterfly flapping its wings in South America, caused a storm in North America.
This “storm” can be said to be a re-planning of Apple’s iPhone 14 series positioning, sales, and other strategies, and it also directly affects the follow-up plans of Apple’s chip team, including products and personnel.
This engineering design error allowed Apple to reorganize the GPU team and adjust some project managers and key personnel in order to solve the defects of the current new GPU and revive the original plan of the A-series chips.
Not only internal adjustments but also part of Apple’s chip development talents began to lose. According to press reports, many of them went to the Silicon Valley chip start-up teams Rivos and Nuvia.
Nuvia and Rivos were also founded by key figures in the former Apple chip team. Among them, William, one of the founders of Nuvia, participated in the development and design of A7~A12X and has always held a leadership role. In addition to the design of the chip, he also served as the layout designer of the A chip in the first few years before founding Nuvia.
“In the past few years, the GPU performance of A-series chips has been limited. This is mainly due to the optimization of the original architecture, not a redesign. Since William left, the CPU of Apple’s A-series chips Performance growth has slowed significantly.”
Recently, Nuvia was acquired by Qualcomm and began to become a springboard for Qualcomm’s self-developed Arm architecture chips. In other words, part of Apple’s core team was poached by Qualcomm.
Rivos, established in 2021, also mainly attracts many Apple chip engineers to join. And Apple claims that these engineers copied many engineering project files when they left, and then provided them to Rivos. Both of them are currently in related litigation.
Nuvia and Rivos are actually just two representative companies that lost Apple chip engineers. Since 2019, the Apple chip team has lost dozens of key engineers to other chip design companies.
Faced with the recent situation, Apple executives have also begun to talk to working engineers one after another, convincing them that Apple’s work has higher value and more stability, “much better than those start-up companies that have to take huge risks.”
Apple is also convinced that the general environment of layoffs in the economic downturn will keep more engineers at Apple.
However, what’s interesting is that when Jobs dug people around to form Apple’s core-making team more than ten years ago, he knew that money could not impress these chip tycoons, and instead expressed the hope that they would complete a historical achievement together with Apple. , which is the challenge of creating a top-level chip from scratch.
Subsequently, Apple gathered a luxurious core-making team in Silicon Valley at that time, and finally successfully fulfilled its wish at that time ten years later.
And the historic challenge that Jobs said has been completed, and the unbreakable cohesion of Apple’s core-making team seems to have begun to loosen. At this time, Johny Srouji, the helm of Apple chips, also needs to make a new plan to revive the core-making goal.
At present, the intergenerational performance improvement of Apple’s self-developed chips is still higher than market expectations. Even though the performance growth of recent generations of chips has slowed down significantly, Apple’s A-series and M-series chips are still not to be ignored exist.
Especially in terms of energy efficiency, it is still ahead of its competitors of the same generation, but this “gap” is gradually being smoothed out. Given Apple’s recent personnel changes, can they maintain their original Some advantages are still unknown.
In fact, it’s not just the A-series chips that have made “mistakes” recently. Before that, there were compromises on the top M Extreme chips.
At the same time, the delay of a series of chips such as M2 Pro and M2 Max is also an impact of Apple’s core-making team’s continuous self-adjustment.
With the successful mass production of TSMC’s 3nm process, Apple’s M2 Pro and M2 Max may be in the first mass production. The same generation of M chips adopts different processes. Perhaps this is very likely for the Apple core-making team. become another milestone.
But this premise is that the 3nm M2 Pro and M2 Max have enough outstanding performance, not just a thorough understanding of the process bonus.