Technology and chip companies

T + T – normal size

Steve Jobs passed away more than a decade ago, but his genius is still high in the technology sector. The co-founder of Apple taught other drivers.

No matter what goes on inside a smartphone case or under the lid of a computer, the only thing buyers are interested in are products and experiences he likes to call “magical”. It was therefore surprising to see that Apple attaches so much importance to the technical performance of its recently designed internal chips. The company first came out with a design for its unique iPad processor more than a decade ago, and its silicone ingenuity has been at the forefront since 2020, when the company launched the Intel processors inside the Mac with its M1 chips. replaced. .

Since then, the company has achieved performance gains that have outperformed competitors. Two larger versions of the M1 were later adopted, and this week Apple combined two of its largest M1 chips to power a new workstation aimed at professional video editing.

While the specifications of these chips may cause boredom for the average customer, these in-house designed processors have become central to the ambitions of major technology companies, and include the “TPU” processor for Google, which is specified using an application artificial intelligence accelerator, and the data center chips “Graviton”. From Amazon Net. Given the dimensions at which such companies operate, their in-house designed chips could predict a broader reorganization of the industry.

One explanation for the new focus on chip designs internally is the amount of chips that big technology companies are eating. Another explanation is that small improvements in the price / performance comparison, for more demanding applications, can yield significant gains. Google’s TPUs, designed to handle data-intensive machine learning, have become the most important in ASICs designed for specific tasks. Similarly, Tesla’s massive funding of AI chips could one day give the company an advantage in self-driving vehicles.

Meanwhile, the complex supply chain for the chip industry has evolved to facilitate access for such companies. For Apple, this meant using designs from technology provider ARM as building blocks for its chips, and on the other hand, taking advantage of Taiwan Semiconductor Manufacturing Corporation’s most advanced production lines to manufacture processors with world-leading specifications.

The ability of technology companies to integrate chip design into their broader technology development plans will give them the most significant benefit and in turn pose the greatest challenge for traditional chip makers.

This merger can take many forms. Cloud computing services can bring new chips from its Annapurna design labs to manufacturing once they are manufactured, and Apple has reported that all of its applications work faster on its own chips.

Not too long ago, Google took a look at the many benefits of designing its own internal chips, with an outline of 10 lessons it has learned to build its own TPU since 2015. It ranges from economical (importance lies in the full cost of the system rather than just chips) to highly technical (maximizing chips to the latest technology possible in neural network design).

With an increase less than the shrinking feature sizes in chips, according to Google researchers, the best hope for new leaps forward lies in speed and performance in the joint design of hardware, software and neural networks, something that takes precedence over power of companies doing all the work.That things are under one roof.

To date, the chip industry’s biggest concern has been central processing units, which have accelerated the penetration of ARM designs into markets that have long been monopolized by the Intel X86 architecture.

The next company to face an awkward situation could be Nvidia, which has become the world’s most valuable disk company thanks to its GPUs, which were initially developed for video games but are now widely used in “accelerated computing applications” such as machine learning.

This week, Apple favorably compared its latest M1 chips to Nvidia’s high-end GPUs. AWS has also claimed significant performance gains over Nvidia from its second-generation Trinium chip.

The critical test of success will be whether the new chips can move large technology companies to new markets. For Apple, it could include self-driving vehicles and headsets for digital and augmented reality. These commodities require a big leap in energy processing. But if the spirit of Steve Jobs is still alive at Apple’s headquarters, those chips will be the last thing on people’s minds when the day comes when they appear in public.

Translation: Noha Hawa


Leave a Comment