In the current semiconductor industry, competition between manufacturers is in full swing. The well-known TSMC and Samsung have quickly taken the lead, announcing the realization of 3nm process technology. However, just as they are feverishly climbing the peak of technology, we see Intel, which has been sticking to the 7nm technology line. Meanwhile, other wafer fabs compete on a variety of technology platforms: 14nm, 28nm, and even 60nm. Does such a technological gap mean a difference in product quality? Or is there some other secret hidden in it?
From the perspective of digital logic, the technical process of 3nm should exceed 5nm, and 5nm should exceed 7nm. Smaller nanocounts usually represent higher density transistors, higher processing power, and lower power consumption. From this point of view, the smaller the number of processes, the more advanced and advanced the technology. Simply put, just like the more horsepower a car has, the more powerful and faster it is.
Nowadays, whenever a new chip comes out, it seems that everyone's first thought is to explore the process technology behind it. That number seems to have become the gold standard for measuring the quality of chips. Take the Kirin 9000S, which has attracted much attention recently, as an example, once it was released, it was hotly discussed on the Internet. People wonder if it's 14nm or 7nm, or 5nm? This pursuit seems to be a process of searching for the truth, trying to interpret the level of the product from behind the numbers.
But in fact, there are some details that have been overlooked. Starting from the 28nm technical process, simple nanodigits can no longer directly represent the physical size of the chip. Many consumers may not know that when we say that a chip is 14nm or 7nm, we are actually talking about an equivalent process.
What does "equivalent" mean? It means a 14nm chip, which isn't actually a 14nm size. This "14nm" is just an identifier, representing that the performance and power consumption of this chip are comparable to real 14nm size chips. Similarly, 7nm, 5nm, 3nm, these seemingly smaller and smaller numbers, are also equivalent technical processes behind them. This may explain why chips of different processes do not have such a big gap in performance, because the actual meaning behind the numbers has long gone beyond the surface.
In the era of rapid development of information technology, every chip is a witness of the times. In the early days, when we referred to "how many nanometers" chips, such as 90nm or 45nm, it was not a random number. It directly reflects the actual length of the gate in the transistor. In the professional world, we call it Gate Length. Just like the genetic sequence in human DNA, every reduction is the ultimate pursuit of performance and the footprint of scientific and technological progress.
Therefore, fabs are like artists who shine ink, constantly trying to reduce the length of this gate on the canvas of technology. But, as with the challenge every artist faces, fabs have found that it's not an infinitely scaled process. As the gate length decreases, the chip begins to exhibit a short channel effect, which means problems such as leakage, instability, increased power consumption, and heat generation. These problems seem to be a test of technological progress, limiting further miniaturization of chips.
In particular, as fabs bravely crossed the 28nm threshold, they found it increasingly difficult to reduce gate length by 30% per generation using traditional methods. Just like a runner in the last section of a marathon, each step requires more effort. The original Moore's Law no longer seems to be as easy to follow as it used to be. However, the fab did not give up, and they were determined to find new ways to move forward.
Faced with this dilemma, fabs have changed their strategy. They began to no longer simply pursue miniaturization, but turned their attention to the three directions of performance, power consumption and area. These three directions have also become their future development goals. Just like a warrior, on different battlefields, different strategies are required. In this technological battlefield, fabs decided to use performance, power consumption and area to confront the challenge.
In the end, the fab achieved some remarkable results. Although the gate length is not really reduced by 30%, if they can achieve a 30% reduction in transistors, or a 30% reduction in power consumption, or a 30% increase in performance, everyone will embrace that this is a new generation of processes. Just as disruptive innovation in every historical period is not simply linear progression. This change in the fab may be a new milestone in the development of science and technology.
In the progress of science and technology, people often pursue a certain visible and perceptible standard to determine whether a technology is progressive. Just as we strive for speed and time in running races, gate length has become such a benchmark in the semiconductor industry. But in reality, technological advances are not always linear and not always obvious. Metrics like gate length, in many cases, seem to stay in place, without any significant improvement. But this does not mean that semiconductor manufacturers have stopped their efforts. Instead, they quietly improved other processes behind their backs, which not only improved the performance of the chip, but also succeeded in reducing power consumption. Such advances may not be as intuitive as numbers, but the effects are real.
As a world-renowned semiconductor company, Intel's investment and determination in technology research and development are undoubted. But compared with other companies such as Samsung and TSMC, Intel is slightly "clumsy" in its publicity strategy. These companies are adept at "exaggerating" their results, claiming that their transistor size has been drastically reduced and performance has been greatly improved. But in fact, although Intel's technology has also improved, it has been polished on the 14nm process for up to four generations, making the outside world mistakenly think that its process is delayed. Meanwhile, TSMC and Samsung have claimed to be moving into the 5nm process, putting Intel at a disadvantage in the digital war.
But from a more microscopic, substantive perspective, the digital war for the process is not as simple as it seems. Using transistor density as a criterion, Intel's 7nm process is actually comparable to TSMC's 5nm process and even Samsung's 3nm process. This means that although Intel claims to be using the 7nm process, its technical strength is no worse than that of other companies. Instead, Intel seems to be "behind" in the eyes of the public by choosing a more conservative number, but this is a misconception because the numbers don't always fully reflect the true face of the technology.
Therefore, when we face today's chip process, we really don't have to be too obsessed with the meaning behind these numbers. Whether it's 7nm, 5nm, or 3nm, these are just words that some companies choose to use to describe their technology, more often than not, they are just a marketing strategy. After all, everyone likes to hear smaller numbers because they often mean more powerful technology. But this is just an illusion, the real technical content, we need to understand and study deeply.
Of course, this does not mean that the numbers are meaningless, but it is more important that we see the essence behind the numbers. The performance of the chip, the density of transistors, the power consumption, etc., these are the real key criteria that determine the quality of a chip. These substantial indicators are far more important than a simple nanometer value. Therefore, when we evaluate a chip, we should not only look at its process figures, but should pay more attention to its actual performance and performance. This is the real measure of technological progress.