With the help of SPARC T2 in the compatible server profit in the company and the digital chip DM series to make money in the Gaosen Semiconductor, with the good development trend and leading advantages in their respective fields, finally at the end of 2010 to get the support of the science and technology innovation fund.

After each company got a billion investment, the equity was firmly controlled in the hands of state-owned assets, even if Liu Ruoyi's investment of one billion funds originally occupied 60% of the shares, after the new funds came in, Liu Ruoyi's shares also retreated to 40% and lost the controlling stake.

Of course, this change does not have much impact on Liu Ruoyi, as long as Gaosen Semiconductor is on the right track and makes no money, Liu Ruoyi really doesn't care about the number of shares, and there is no shortage of this in her industry.

For Zhao Mengguo, Li Shijie and Wu Zuhua, equity changes are not a thing, everyone knows that after this company is done, the controlling stake will definitely be put into the hands of state-owned assets, and the important thing is that now there is finally money in hand.

Zhao Mengguo's research on improved chips based on SPARC T2 was originally in full swing, but now with more abundant financial support, Zhao Mengguo also carried out project research on the main hardware around SPARC, such as motherboards, such as liquid cooling solutions.

These things are the core components of the future independence, with the current SPARC ecosystem, in the short term, there will definitely be no manufacturers to do this matching, the market is too small, this kind of highly specialized things must come by themselves.

On Li Shijie's side, after gnawing down the 64-way super server, he is no longer satisfied with the small fights in the field of small computers.

Looking at the excellent quality of this batch of SPARC T2s, Li Shijie pondered for a long time, and made up his mind to pull a unit that had done supercomputer research and development, and began to set up a project to develop a compact supercomputer, which is sometimes called HPC.

Of course, he is not going to be the kind of large-scale supercomputer that is leading in the world, but mainly for practical purposes, developing affordable and streamlined supercomputers to make up for the computing power gap between super minicomputers and large-scale supercomputers.

The cost of this model is much lower than that of large supercomputers, but it is far superior to super minicomputers in terms of computing power, and is especially suitable for applications with high computing power requirements.

The initial goal of the project was to build a modular dual-CPU computing unit to build an HPC system, which is like combining two CPUs into one, and then plugging and dialing 128 computing units to combine into a compact supercomputer.

In this architecture, the size of the 128-block computing unit model may only be the size of two cabinets, and the future operating system will still be mainly running the Solaris system, and will be used as a 128-channel super server. After all, SMP systems usually hold up to 128 processors, and if you want to make some fuss about this architecture, you can only do some workarounds.

In the future, after the design capability of the computing unit is improved, the computing unit can also be designed as a computing unit with 4 CPUs, and a new 128-channel server can be designed, but in fact, 512 SPARC T2 CPUs have been integrated into it.

This kind of supercomputer, which actually integrates 256 CPUs or 512 CPUs as computing power, may cost more than 10 million yuan, but in terms of performance, it has increased the high-end computing power by several times compared with the 64-channel super server, relatively speaking, it has a good cost performance for users who need to be scientific computing floating-point operation-intensive.

After all, this is a new product with most of the performance of a supercomputer, but it may cost less than a tenth of the price of a supercomputer.

What's more, the product will continue to be compatible with most of the previously developed Solaris applications, and the environment will not be as demanding as that of supercomputers, and the system will be much cheaper to run, which is an advantage that other supercomputers can't get.

Sun Microsystems has rich design experience and products in HPC, of course, the current 64-channel design based on SPARC is a top technical secret that is subject to control, and it will not be open sourced now, and many of these HPC servers used to be restricted products for the mainland.

If Li Shijie's plan can be successful, then in the high-end computing power market, such as weather forecasting, geological exploration simulation computing, and even in various fields such as data center main servers, Shenwei thin supercomputers can occupy the commanding heights of technology.

This modular combined supercomputer also retains good flexibility in future upgrades, and may even support a new model of computing unit, which is a simple way to upgrade chips.

Moreover, after accumulating experience, we really need to study large-scale supercomputers in the future, and we also have a good foundation, and the way to use multiple sets of HPC to form large-scale supercomputers also exists from the perspective of technological evolution.

The SPARC T2 has similar features and performance to the mainstream PowerPC processors, and has developed so many world-class supercomputers under the PowerPC architecture, and based on the SPARC processor, Toyo FJT has also played a role in developing the world's leading supercomputers.

Li Shijie believes that Shenwei can also tap this potential on the basis of SPARC T2, and it is very possible for Shenwei to develop a supercomputer with the world's top 500 performance in the future.

It's just that now their technical capabilities and financial resources do not support this kind of research and development work.

Without the support of state funds, the current large fund funding is not enough to toss around chips and large supercomputers at the same time.

At present, it is best to borrow a boat to go to sea, while developing a streamlined supercomputer to accumulate technical experience, and at the same time, you can also use the product to commercialize and make money, and then further support new research and development work.

That's the way to go the long way.

Li Shijie, who has tasted the sweetness of commercialization, is now even more interested in the development of business than in terms of technology.

After being poor for so many years, it was hard to take this opportunity to find a new way, and Li Shijie certainly knew that the right path of commercialization could support them to go further.

The original approach of only the advancement of scientific and technological achievements has not been combined with the actual market demand, and most of the time it has been useless and cannot be transformed into actual products, and then it lacks the necessary hematopoietic function in follow-up research and development to support sustainable development.

Now the convergence of the two institutes is not so much a combination of two benefits, in fact, it is more about huddling together for heating and using limited resources to find a new way.

In fact, the restructuring of scientific research institutes under the conditions of market economy to the market is very unsuitable for most people who have studied technology all their lives.

In this way, many technical talents are drowned in the tide of the market economy.

Whether it is Li Shijie or Zhao Mengguo, this time he took the opportunity to rise when he encountered a good opportunity.

And Wu Zuhua, if he hadn't met Liu Ruoyi, he would have worked hard for a few more years and waited for more policy support before he could barely speak.

Their work is much better now than on another plane.

At least, now they all have a hematopoietic industry with an annual profit of not less than 100 million yuan to support.

Gaosen Semiconductor's 3A-1500 has been finalized, and from the current test situation, all indicators are very good, with low power consumption and thermal power, and high reliability.

The use of this independent product in the fields of aviation, aerospace and navigation is self-evident. Now, after the processor comes out, Gaosen Semiconductor has further launched a package of turnkey solutions such as supporting motherboards and operating systems.

The first batch of 500,000 pieces of OEM orders has been handed over to the Dragon factory for 45nm process production.

The control of all aspects under this new process is relatively ideal, and the cost is relatively moderate, and Gaosen Semiconductor has finally solved the problem of blank field of high-performance and low-power processors in the mainland this time.

For the high-performance embedded autonomous processing chip that is urgently needed for intelligence in the mainland, it is finally said that there is a very satisfactory main chip to support the development of new applications.

The 3A-1500 chip can be set to run at a single-core 1500MHz main frequency, up to 4-core 2000Mhz frequency high-performance full-load operation, respectively, the operating parameters can be set in the system, if the heat dissipation environment is harsh and the temperature is high, the system will automatically reduce the frequency to a single-core 1500MHz to reduce thermal power consumption, and exchange lower performance for reliability in a high thermal environment.

For increasing the reliability and adaptability of the system, this 3A-1500 chip from Gaosen Semiconductor already has a very good solution.

With the increasing number of 3A shipments, many systems that originally used PowerPC603/604, a low-end and high-reliability chip from various sources, have now begun to upgrade and transform their research and development work.

In terms of computing power, it may take a few minutes to get the result, but now the performance is greatly improved after changing to a new processor, and the computing function may be completed in a few seconds, and the 4-core 3A chip can easily form a variety of embedded high-performance computing systems such as dual CPU single boards and 4 CPU single boards.

Now this 3A computing power is compared with the 603/604 chip of the last century, which is exactly like the product of two worlds.

The combination of 8-channel and 16-channel single board has been in full swing in research and development, which is the responsibility of Gaosen Semiconductor, and when the first batch of reference solutions are released, the fully autonomous and controllable low-power, high-performance and high-reliability computing platform will have a huge increase in computing power.

The significance of this computing platform cannot be measured simply by money.

It's just that for Liu Ruoyi, a major shareholder and board member, he can't see these things at all.

The R&D system of the company and the institute is completely separated, and only those who are willing to make public the projects on Wu Zuhua's side will have a progress briefing within the company, and if Wu Zuhua feels that it is not necessary, there is only one project on the company's side.

This practice, of course, is not a fully commercial operation.

Liu Ruoyi knew it, but pretended not to know anything.

In the years before she crossed back, Liu Ruoyi, who has been a little researched on semiconductors, is of course inseparable from the various breakthroughs of so many semiconductor companies in the mainland at that time.

The processor is autonomous, the memory is also autonomous, and the energy-efficient DSP is not far from autonomy if the design is more strenuous and the various key analog components are popping up a large number of manufacturers...

Now that these are iterating faster than the previous time and space, the butterfly effect of Liu Ruoyi's entry into the semiconductor and technology world is finally beginning to show great effects.

......

Now Liu Ruoyi has stirred up the semiconductor pattern of the world, at least by speeding up the process of the original mainland for more than five years, and some of the results are even more than the results of the original time and space ten years later.

For the world, it is not even the changes in the business landscape that can be seen now, but also the deeper changes that people in the outside world do not feel at all.

The mass production of the 32nm process technology at the Dragon Factory is, in a sense, an important milestone breakthrough in the development of semiconductor technology in the mainland.

With this thing, Liu Ruoyi's chip research and development, and the players of the national team in domestic chip design and development, have an extremely important support.

Otherwise, if the chip is handed over to the 32nm process foundry of the island or any other foundry, it means that all technical secrets must be opened to others, and this so-called independent and controllable is simply a joke.

However, if you don't open up most of the core design data to the foundry, you can't produce this thing at all.

And after the core information of the hardware design is disclosed, the confidentiality value of this thing is greatly reduced.

This is the demand of the mainland in special application fields, obviously there is a more powerful SAPRC T2, but it is basically mainly used in the intermediate research and development process of computing power support and private commercial applications.

On the master equipment terminal, the relevant departments prefer to use SPARC performance ratio T2 is still a lot short of Gaosen Semiconductor 3A-1500, it is because Gaosen Semiconductor 3A-1500 has completely redesigned itself from beginning to end, although the main computing core has not changed much, but it has added a lot of highly confidential hardware encryption circuits to strengthen the security and confidentiality of the entire hardware, including hardware encryption and decryption circuits, which not only improves the speed of encryption and decryption, but also enhances the resistance to external intrusion from the source.

If it weren't for some of these considerations, given the existence of electronic warfare and cyber hacking, using a device with someone else's hardware and operating system installed would be as dangerous as sticking your neck under someone else's knife.

At this time, with Liu Ruoyi's rapid development in the semiconductor industry due to fate, the relevant special departments can also be regarded as borrowing ships to go to sea, taking advantage of this shareholder wind to set off a wave of intelligent upgrading of the main equipment, 3A-1500Although it is still a little worse than the current mainstream processor capacity, it is an order of magnitude improvement compared to the computing performance used before.

Now the most important thing for everyone is to verify its reliability and stability, and once it is confirmed that these are not problems, it is the beginning of large-scale application in various large-scale equipment.

The key components such as independent processors and memories have the reliable processing power of Zhongxin International to borrow.

In this regard, although the equity of Zhongxin International has been very private, in terms of core management, Zhang Rujin is only controlling the overall situation in terms of R&D and business development.

In terms of specific production, it is not surprising that some production lines have mysterious orders with a high degree of secrecy, and the relevant technical information will be withdrawn after production.

Especially on the 45nm process production line not long after the completion of the DreamWorks headquarters in the mainland, this kind of business is increasing.

As for this situation, Zhang Rujin just pretended not to know, anyway, there will never be any rumors about this matter outside, the production line is controlled by his own people, and it is much more convenient to do anything.

As always, the Dragon factory is introducing the latest equipment for the R&D team to break through the 28nm process and research the 22nm process, and the current mass production of the 32nm process technology is only the first goal to keep up with the world's semiconductor foundry team.

Now only in the 28nm process technology and 22nm process technology with the opponent or even ahead, Zhongxin International can be said to be really in the same running line with the competitors.

The opponents are still maintaining the advantage of leading half a generation, and the yield rate of the 32nm process process that has been mass-produced by Zhongxin International in one year ahead of schedule has long been more than 90%, and now their high-end capacity utilization rate is saturated, and the high-end process process production line has contributed more than half of the profits to TMS.

The competitor's most advanced 28nm process process has been successfully tested, and it is no surprise that mass production will be carried out in 2011, and the yield problem is now being solved.

If Q1 in 2011 can be successfully tested, the dragon factory is expected to catch up with the 28nm process only 2 to 3 quarters behind.

It can be said that the battle of 28nm process technology is the most important battle for Dragon Factory to enter the international OEM market.

From a technical point of view, the technical gap between the 45nm process and the 32nm process is actually not that big.

But the progress from 32nm to 28nm is a singularity that has increased dramatically. From the perspective of equipment investment alone, the 28nm process production line will start at 5 billion US dollars, and the research and development will burn Liu Ruoyi at least nearly 78 billion US dollars.

Waiting for the relevant equipment to return to its original price after the technology matures in the future, a 28nm process production line with a monthly production capacity of 30,000 to 50,000 wafers is expected to reduce the equipment purchase cost to about 3.5 billion US dollars.

Such an expensive production line, of course, is also very expensive in use, at present, if a new chip is to be taped out in the 28nm process, the cost of each tape-out will start at least two or three million US dollars!

Therefore, only the owner of a big bully like Apple dares to ask for free tape-out!

And Liu Ruoyi's start-ups like Blue Core Technology can only find 130nm process technology in the mainland to tape out.

Bluetooth chips, a low-cost chip, cannot afford to use a 32nm high-end process production line.