On the eve of the release of new generations of processes and video cards from AMD and NVIDIA, it is worth examining such an important characteristic of the chip as the technological process of its production. Intel has been riveting 14 nm processors since 2015, while AMD and NVIDA have been using the outdated 28 nm process technology. From our article you will learn about what the chip manufacturing process is and how it affects the main characteristics of the CPU / GPU, as well as the answer to the question: “Which is better: buy now or wait for the next generation?
When announcing video adapters Pascal (NVIDIA), Polaris and Zen (AMD) all major semiconductor manufacturers have announced their transition to a new manufacturing process using FinFET transistors.
Intel has been using all these innovations since the Broadwell generation in 2015 and does not plan to change them to a thinner 10 nm process technology in the next generation after Skylake, by the way, you can read about Intel technologies here.
AMD has chosen 14nm from GlobalFoundries and Samsung for its Polaris GPUs and CPU Zen, which is less than TSMC’s 16nm from NVIDIA. And you can read about the technologies of these companies by following the corresponding links: AMD, NVIDIA.
It should be noted in advance that all the subtleties of the production of transistors will not be touched upon here, here you will simply learn about the meaning of a more subtle technical process.
What is a technical process?
In general, the technical process for the production of semiconductor circuits implies a sequence of various technological and control operations. But why then in the graph of the technical process is written a number with the designation in nanometers? It’s just that the photolithographic equipment, with the help of which transistors are obtained, has a resolution. To better understand this, we advise you to watch this video:
Over time, there is an evolutionary improvement of this process, which allows you to still observe Moore’s Law.
Interesting fact: Intel Pentium had a 800 nm process technology, which seems like an insanely large figure by modern standards! And only 3.1 million transistors. (Intel Core i7-5960X has 14 nm and 2.6 billion transistors)
What does the process technology affect?
It is not for nothing that manufacturers are proud of the new achieved level of this technological process. After all, it provides tangible benefits:
- a decrease in the transistors themselves leads to an increase in their number per unit area, and this increase allows either placing a larger number of transistors on the substrate, which increases performance by expanding the number of computing units or reducing the area of the substrate itself while maintaining the same number of transistors.
- the smaller size of the transistors reduces their heat and power consumption. This allows you to either increase the frequency and the number of computing cores without compromising heat dissipation or simply reduce power consumption, which is especially convenient for laptops.
- FinFET transistors are often used along with 14nm process technology. These are transistors that have a three-dimensional, fin-shaped gate that can reduce the size of the transistor and reduce current loss and delay. There are several types of them, but they will not be told about them here, so if you are interested, then go here.
- the transition to a new technical process requires new equipment, which is an expensive operation. This primarily affects the price of processors.
- the transition to a new stage does not occur immediately. The technology needs to be tested, so the first chips on a new technological process may not be produced the first time (affects the price). Especially this complexity grows with an increase in the chip area, which does not allow immediately after the presentation of a new technical process, immediately “sculpt fast multi-core chips with a huge crystal area. This is more true for top-end video chips, where up to 12 billion transistors can be used!
So what should you expect?
If you think about it, it turns out that a significant leap in energy efficiency should be expected this or next year, which will allow raising the frequency for top-end chips and reducing cooling requirements for cheap ones.
By video cards
In this generation, AMD with their Polaris (RX 400-series) has every chance to compete with NVIDA with their Pascal because their process technology will be less, which can compensate for the increased heat dissipation of cards based on the GCN architecture. Although in addition to the technical process itself, both manufacturers will present a new architecture, which can show us a new level of performance, after all, the 4K standard is gaining momentum. If you want to know some interesting facts about your video card, then you are here.
As for the processors, here AMD promises us a 40% increase in performance per clock, which promises healthy competition with Intel, which have recently become lazy, their 5% increase in Skylake has upset many fans. Also, with such a leap in process technology, Zen can finally give Intel a real boost in energy efficiency. Old 28 nm could not compete in this parameter.
Also at the moment it is already known that Zen processors will not replace the FX and Opteron, these chips will not be released beyond 2016.
There are high hopes for the Zen microarchitecture, because Jim Keller had a hand in its development. He is known as the developer behind the DEC Alpha 64-bit RISC, which then became the AMD K7. He created the AMD K8 architecture after which he left AMD in 1999. Now, after returning in 2012, he again leaves the Reds.
We ask you to forgive us for such a small digression into history, maybe someone will be interested in this topic.
The chip manufacturing process has a very large impact on parameters such as power consumption, the number of transistors and indirectly affects performance.
In addition to upgrading the technical process, AMD and NVIDIA are also demonstrating new architectures, which together will allow for a leap in energy efficiency and performance.
So if you are tormented by the question of whether it is worth waiting until the new release of new video cards and processors, or buying here and now, we tend to the second option. An exception will probably be the case with the most powerful video cards, since their release may be delayed due to the large area of the chip.