Rebirth 08: Rise from copycat phones

Chapter 304 Advanced Intelligent UAV Combat System

As November approached, a piece of news of some note appeared in the supercomputing circle.

The super-large computing center "Zhisuan No. 2" invested and built in Shenzhen by Zhiyun Cloud Computing Co., Ltd., a subsidiary of Zhiyun Group, and Liming Information Co., Ltd. has successfully completed various preliminary tests and was officially put into use.

What is amazing is that once Zhisuan-2 was built, it became the world's number one with a floating-point computing speed of 10.3 billion PFlopsc, far surpassing the second-ranked Tianhe-2.

The floating-point computing speed of Tianhe-2 is only 33.86 PFlops.

This is not the point!
The key point is that all the chips in Zhisuan No. 3 are the same as those in Zhisuan No. 6, all of which use Zhiyun's self-developed chips, including the new generation of SOP instruction set server chip architecture ZS1100A-SOP chip.

The chip adopts a new architecture ZS3 and has extremely powerful performance.

The ZS3 series chips are also the third-generation server chip architecture of the Zhiyun Semiconductor SOP business unit under Zhiyun Group, and are specially designed for high-performance servers.

Compared with the first and second generation SOP server chips, a new generation of architecture and new process using 28NLP process is adopted.

The ZS3A6 chip has eight cores and supports up to sixteen threads. Apart from slightly higher power consumption, its performance can be said to be quite powerful.

In addition to the ZS3A6 chip, Zhisuan No. 1500 also uses the APO computing power card from Zhiyun Semiconductor.

The APO1500 computing power card is the most advanced, top-level, and most professional general-purpose commercial computing power card. Even NVIDIA, which was also acquired by Zhiyun, has several high-end graphics cards on the market that are far inferior to this APO computing power card.

Of course, this comparison does not take into account Zhiyun’s secret project AI series chips, otherwise there would be no way to compare... The performance is many times worse, especially in the special field of AI training, the performance gap can reach more than ten times.

Zhiyun's AI series of chips are still top-secret projects to date. All projects using or involving this series of chips are required to undergo a confidentiality review. The outside world does not even know that such a chip exists, so it naturally cannot be used in regular commercial projects.

The main reason is that the AI ​​chip developed by Zhiyun has terrible performance and a single purpose... This thing has only one purpose, which is to train AI...

Currently, AI technology and core AI chips are subject to extremely strict control.

After all, the core use scenario of AI technology these days is not commercial... Although it has some commercial applications, it is not large.

The main application areas of current AI technology are actually drones or other unmanned mobile terminals.

The secret team under Zhiyun Group is currently cooperating in the development of an AI project for super-maneuverable supersonic stealth drones...

This type of aircraft can not only fly autonomously and execute various instructions, but can also be used in conjunction with manned aircraft.

There is a very fashionable word in the outside world: loyal wingman!
But it is actually a level higher than the loyal wingman rumored by the outside world, because the "advanced intelligent drone combat system" developed by the secret project team of Zhiyun Group does not need the command of manned aircraft in the rear, and it can complete a series of tasks by itself.

Including take-off, path planning, penetration, execution, return and landing!
No remote control is required at all!
A true drone!
It sounds very sci-fi, and it is actually very sci-fi!
Of course, for safety reasons, limited remote control via satellite communications is actually still required, such as confirming the final strike order.

You can't just release an AI and let it fly around randomly... I don't trust it...

This will require the support of a large number of low-orbit satellite communication networks. It is impossible to use a synchronous orbit communication satellite for remote control because the delay is too great.

While the outside world was still pondering whether loyal wingman was feasible and conducting various experiments, Zhiyun Group had already started working on it.

In the beginning, our partners were just talking casually. Some even brought out some concepts from science fiction movies, saying that we can handle the airframe and we can just do it manually, but the intelligent control system is very difficult to handle and cannot be done manually. Do you have any way to do it?

The engineers of the secret team of Zhiyun team thought about it for a while, and then said to them: It’s not impossible...

Then the project “Advanced Intelligent UAV Combat System” started.

Not an aircraft, but a combat system.

This project was directly classified as a top-secret project at the highest level for control... because apart from other technical indicators, the maneuverability index of this aircraft alone is too scary.

When performing maneuvers, the overload can reach fifteen Gs.

And relying on powerful onboard computing power, it can instantly plan the most optimized flight path, which is very scary.

When performing simulation calculations, the aircraft under the control of the system can maintain a stable altitude of only tens of meters above the ground, which is the height of treetops, and can fly at supersonic speeds over undulating terrain...and can also perform high-overload maneuvers during this process, giving the audience a lot of fun.

This kind of extreme flight would make even human pilots speechless: the speed is too fast, the altitude is too low, and the terrain is uneven. At the same time, the air currents in the low-altitude environment are changeable, requiring the aircraft to make various adjustments very promptly and frequently.

Humans simply can’t react fast enough.

It can also dive straight down at high speed at 90 degrees, and when it is finally pulled up, the height from the ground can be controlled within one meter... Note that it can be controlled at one meter every time.

If a human pilot dares to do this, the plane will be destroyed and the pilot will die out of every 10,000 attempts.

But AI can, and it can control the altitude at one meter every time. Unless there is a malfunction in the aircraft, there is no such thing as failure.

This is a reaction speed, control accuracy and overload tolerance that humans can never match!

As long as the aircraft is strong enough, the various sensing devices are advanced enough to collect enough information, and the computing chip is powerful enough, AI can theoretically achieve any extreme flight that only exists in theory... After all, for AI, it's just a bunch of data analysis, judgment, and execution.

This is the charm of applying AI technology in the field of drones, and it is also the reason why Americans in the original time and space were determined to ban AI chips.

Similarly, this is also the reason why Zhiyun Group’s AI chips are strictly controlled.

At least before major technological breakthroughs are achieved overseas, especially in the United States, don't expect Zhiyun to announce any news about AI chips.

However, AI technology cannot be commercialized, and commercialization also requires some computing power card chips, so this task basically falls on the super castrated version of the AI ​​chip... the APO chip.

The Zhisuan No. 3 supercomputer uses the APO second-generation chip and the ZS6A chip with the SOP instruction set.

Then it obtained the world's number one supercomputing power ranking!
In addition to providing supercomputing power for various projects under the group, the supercomputer is also open to the outside world and accepts various commercial computing power orders. It is mainly provided to universities, research institutes, and corporate R&D institutions, and mainly serves scientific research.

It doesn't have much to do with ordinary people, so it doesn't attract much attention among ordinary people.

Xu Shenxue himself is more concerned about this. He personally doesn't know much about supercomputers and doesn't have much interest in them. However, the group of research and development projects under him require a large number of supercomputers!
This is an urgent need for many R&D projects of the group!

Previously, many projects used the group’s own internal supercomputer, namely Zhisuan No. 1. However, the performance of this supercomputer was only average, and sometimes it was difficult to meet the huge computing power requirements of a large number of the group’s projects. Many projects had to go to other supercomputing centers in other cities.

Now with the Zhisuan No. 2, many computing power needs within the group have been basically met, and there is still some surplus to be supplied to other scientific research units.

------

In late November, Xu Shenxue once again appeared at Zhiyun Microelectronics' Shenzhen Semiconductor Base.

This time he did not come to see chip manufacturing or advanced process breakthroughs, but to see the small-scale trial production of the country's first DUV immersion lithography machine.

"It's a pity that our lithography machines are still subject to some patent restrictions, otherwise we could export them and grab some orders from ASML!"

Wang Daolin of Gulf Technology said with a look of regret: "Now, it can only be used in our country!"

Xu Shenxue said: "What's so regrettable about this? This kind of thing needs to solve the problem. As long as it can be solved, then it's not a big problem!"

"Moreover, our lithography machines have complete patents in China. The most we can do is not be exported!"

For a national important equipment like lithography machine, it is not easy to apply for various so-called independent patents abroad, after all, ASML, Nikon, and Canon have already been there, and many patent applications cannot be applied for. But in China... Gulf Technology's patent applications have been green-lighted all the way, so at least in the domestic market, Gulf Technology's several lithography machines are all 100% independent patented technologies, and there is no question of infringing ASML's patented technology. At most, the technical features are similar...

Of course, anyone with a discerning eye knows that Zhiyun’s DUV immersion lithography machine is basically a copy of ASML’s lithography machine, and has infringed on a large number of ASML’s patents.

It’s just that although everyone knows about this kind of thing, they all tacitly agree not to talk about it... Even ASML is too lazy to say it, because it’s useless to say it.

Could it be that ASML can still count on the domestic market to ban Gulf Technology's DUV immersion lithography machine just because of a patent issue?
Even in heaven you can’t dream such a beautiful dream!

If you have this fantasy, foreign countries might as well sue domestic warships, aircraft, tanks and the like for infringing on the patents of similar foreign equipment.

To be honest, the possibility of winning a lawsuit against airplanes, warships and tanks for patent infringement is slightly greater than the possibility of winning a lawsuit against Gulf Technology's DUV immersion lithography machine for patent infringement.

Of course, even if it has so-called "independent patents", Gulf Technology will not shout about it everywhere... technical personnel still have some self-respect.

No matter how much they try to cover it up and fool outsiders, everyone knows in their hearts that this thing is copied from ASML's lithography machine.

At this time, Wang Daolin said: "However, we are also continuing to carry out technological research and development, striving to bypass these patent restrictions as soon as possible so that our lithography machines can appear in the public eye. When the time comes, we will promote our DUV immersion lithography machines overseas and export them to earn foreign exchange!"

Xu Shenxue said with a smile: "It's good to have this confidence."

While Wang Daolin was talking to Xu Shenxue, the first domestically-produced DUV immersion lithography machine was also undergoing trial production in the dust-free workshop in front.

Ding Chengjun beside him was also talking about the trial production plan this time: "The chip we are currently trial-producing is a logic chip in the group that uses a 40-nanometer process. The purpose of this trial production is mainly to look at performance indicators such as yield and efficiency. As for the lithography accuracy, we have conducted a lot of tests before, and it is fully capable of meeting the accuracy of our 40-nanometer lithography machine!"

"Our lithography machines also have the overlay accuracy required for the 28-nanometer process, but they are still lacking in some technical indicators. If we use the 28-nanometer process at this stage, the production cost will increase significantly."

"Even in theory, our lithography machine can barely be used for 18-nanometer or even 14-nanometer production, but the cost is too high and it has no actual commercial value!"

"At present, our microelectronics plan is that the HDUV-400 model purchased in the early stage will mainly be used for the production of logic chips with , , or even nanometer processes, or some storage chips with , , or nanometer processes!"

“We have a lot of orders for these process nodes, and the demand for production capacity is huge, which is enough to give us the starting orders for the 400-type lithography machines!”

“But the prerequisite is that it can meet the performance requirements. It is not just about being usable, but it must be mature and can be used on a large scale, with controllable costs and commercial value!”

At this time, Wang Daolin said: "We are actively conducting technical research and development to solve some of the precision and cost control problems. This is why we need to bring the equipment to your microelectronics. We need a lot of testing and cooperation from the wafer factory before we can solve the problem, improve the yield, and control costs!"

Ding Chengjun said: "Don't worry about the coordination issue. We will actively provide various technical support."

Wang Daolin said: "Give us three months, and we will be able to solve these problems. It will definitely be able to meet the needs of the 28-nanometer production line. Mr. Ding, please place a few more orders!"

Ding Chengjun said: "As long as you can solve the problem, there are plenty of orders, so don't worry."

Wang Daolin: "Then it's a deal. You have to give me an order of at least 40 units for this batch. Otherwise, I will be under too much pressure. The research and development of these lithography machines is burning money all the time. I really can't bear it without orders!"

Ding Chengjun glanced at him and said, "How dare you ask that? A DUV immersion lithography machine costs tens of millions of dollars, and you want an order for 40 machines right away!"

After that, Ding Chengjun added: "However, as long as you can solve the technical problems, I still say that there are plenty of orders, and 40 units is not a problem!"

"This market is big enough to accommodate your order of forty units!"

"However, your 400 model is still a little bit inferior. The overlay accuracy is not very good, only nanometers. It is okay for the -nanometer process node, but it is not enough for our -nanometer process or even -nanometer process!"

"Once you have improved the overlay accuracy and done double exposure well, and are able to produce 18 nanometers, 14 nanometers, or even 7 nanometers, are you still worried about not getting orders? By then, all my orders for microelectronics lithography machines will be yours!"

Wang Daolin said: "This is unfortunate. The 500 model we are developing is just right. The overlay accuracy has been greatly improved to nanometers, and the double exposure technology has also been upgraded and improved. This level is enough for mass production of -nanometer chips, and the production efficiency is very high and the cost is low!"

"How about you place a few pre-orders first? Don't worry, I'll be able to supply you with the goods by next summer at the latest!"

Ding Chengjun was speechless: "At least you can produce a prototype and then come back to get an order!"

Xu Shenxue watched from the side without saying much!
In recent years, Gulf Technology has been taking advantage of Zhiyun Microelectronics and trying its best to make money from it, from the early i-line lithography machines, to the KRF lithography machines, then to the DUV dry lithography machines, and then to the current DUV immersion lithography machines, plus some other core equipment such as etching machines, etc.

It hasn't been released yet, but I've already placed a pre-order with Zhiyun Microelectronics.

However, the project team of Gulf Technology is indeed powerful. They basically produce products with guaranteed quality and quantity, so that the investment of Zhiyun Microelectronics will not be wasted.

The various project teams at Gulf Technology have a total of seven or eight S-level scientific research quotas, which are used in rotation in each project group. Their technical capabilities are definitely strong enough.

Today, the entire semiconductor industry, especially the equipment manufacturing and consumables fields, which is under the Xiannvshan Holding system, has occupied almost half of the S-level scientific research quotas in Xu Shenxue's scientific research system, totaling thirty-five.

Including Zhiyun Microelectronics, which is engaged in chip manufacturing, Zhiyun Semiconductor, which is engaged in chip design, and then adding the AI ​​project and various other miscellaneous projects, there are only fifteen S-level scientific research quotas used in the entire Zhiyun Group and many other companies under Xu Shenxue's name.

Except for a few core projects, other project teams have to take turns using the S-level quota.

However, at Xiannvshan Holdings, many projects have S-level scientific research quotas available all year round.

This shows how important the semiconductor equipment and consumables sectors are to Xiannvshan Holdings.

----

While Ding Chengjun and Wang Daolin were talking and chatting with Xu Shenxue, the production test results of the Type 400 lithography machine at the front also came out.

The technicians quickly brought over the corresponding test results, and also attached some technical parameter comparisons with the ASML lithography machine.

Xu Shenxue took a look and felt that it was very good. Although it still lags behind ASML's advanced lithography machines in all aspects, it has reached a usable level overall.

At least there is no problem in using it to produce chips using the 28-nanometer process. It's just that the production cost of this test prototype is relatively high at this stage, and it is not cost-effective to use it directly for commercial production.

However, this is just a test prototype. Houtou Bay Technology will make some technical improvements and adjustments based on the test results of many actual trial productions. When the mass production version is actually produced next year, it will be able to meet the commercial production requirements of the 28-nanometer process.

The Type 500 lithography machine that Gulf Technology is developing has better double exposure technology, which will further enhance the overlay accuracy and can be used for -nanometer and even -nanometer processes.

Even in extreme cases, we can force ourselves to make chips using a seven-nanometer process.

It's just a matter of cost.

The problem of lithography machines has been barely solved so far. Although it is still not very easy to use, it is there at least.

Having this thing at home will give you more confidence when fighting or negotiating with others outside... In the past year, Xu Shenxue has been threatened with supply cuts to core semiconductor chips such as lithography machines at any time.

Although the supply has not been really cut off, Xu Shenxue was still very disgusted.

Now that he has a DUV immersion lithography machine, if anyone dares to say that the lithography machine can strangle him, he will tell the other person to get out.

Xu Shenxue inspected the first DUV immersion lithography machine for trial production testing. Before long, other people heard the news and came over to inspect as well.

Many people are very curious and want to see what this national treasure looks like!
Not long after, the 7:30 evening news officially reported the news of the first generation of domestically produced DUV immersion lithography machines with fully independent intellectual property rights, which immediately caused quite a stir.

(End of this chapter)