Liu Ruoyi had no idea what she had changed.

She did not attend the celebration of the success of the 32nm process tape-out.

For her, this seems to be a natural thing. I was happy, but I wasn't excited.

She is not Zhang Rujin, and she has a strong competitive complex.

It's not Li Shijie, who has different feelings.

It's not Liang Jisong, who needs to be in a hurry to show his worth.

At her core, she is a unique industry investor.

This is also her old business, invest and reap, that's all. Just like the investment in Wu Zuhua, she has no feelings for Gou Yu, but she feels that Wu Zuhua is a serious person.

Serious people Liu Ruoyi usually think that it should be worth helping, and there are good investment prospects, although it is not clear how much this rate of return will be, Liu Ruoyi still does not hesitate to invest.

Moreover, in order to help solve this matter satisfactorily, the additional source of intellectual property control problems is added.

Liu Ruoyi never wanted to wait for Wu Zuhua to really make successful results here, and then others would suppress their efforts in a state of immortality as soon as they filed a complaint.

So she cleanly won MIPS, and then found that there was an opportunity to enter mobile computing, and did not hesitate to take the opportunity to win Solar Microsystem!

This money is something that can be considered completely pulled through.

With the acquisition of Sun Mobile Computing, Liu Ruoyi is equivalent to mastering a sophisticated R&D team to build a cloud service system for the future mobile computing background.

From the current point of view, Liu Ruoyi's way of building this ecological chain is very reliable.

With these two companies, Liu Ruoyi's semiconductor empire began to take shape.

Mobile computing's terminal computing, storage, cloud virtualization, and hardware design have now gradually formed a chain through Qimonda, MIPS, and Sun Microsystems.

In the future, no one can imagine what wonderful changes will occur in the value of this ecological chain.

It's just that Liu Ruoyi is becoming more and more confident.

Courtyard X.

Li Shijie's team got the first batch of SPARC T2 chips, and he got an old 4-way SPARC T2 server in front of him, which is a high-end server composed of 4 1.2GHz servers produced by 65nm process technology, and now he is replacing the latest 32nm process technology in Jackie Chan factory to produce a new version of the processor.

The palms of the hands at hand were a little sweaty, and it was impossible not to be nervous.

Dismantling, cleaning, reinstalling, inspecting,...

Step by step, the test instrument is ready for temperature and power consumption.

Boot.

...

...

The start is extremely smooth, and the set 1.6GHz operating frequency doesn't know if it's a psychological effect, Li Shijie feels that the start is much faster than before.

There will be a start-up time test result to see if it is an illusion.

There is no problem with the hardware self-test at all.

Enter the system, open the application, select Scientific Computing, import a simulated test data...

...

After 3 minutes, the program outputs the same result, which is exactly the same as the expected result!

"The speed has increased by more than fifty percent, Lao Li, you can!" Song Jianjun, who had been watching the test behind him, couldn't help but praise.

"It's just a change of heart, and it feels completely reborn. This dragon factory is really cowhide. ”

"The single-core frequency has increased by more than 30 percent, a total of 32 cores, this increase in speed does not take into account the optimization of the computing power improvement of the computing platform, if the application is further optimized, I am sure that it will be increased by 100%!"

Theoretically, the performance of the new processor should be increased by more than 2.5 times to be normal!

However, the application is not optimized well, and a lot of the improved computing power is wasted in useless waiting.

"In this way, the future growth space of our main server is really worth looking forward to. "I think that Li Shijie once said that he would use this new processor to develop a new server of 64.

Song Jianjun is now full of fantasies.

The performance of 4 CPUs is already so powerful, what will happen to the performance of 64 CPUs?

"It's a pity that our research hasn't kept up, and the SPARC Open Alliance hasn't opened more than 16-way design guidelines, so we can only work backwards based on the original server of Solar Microsystems, and the progress may not be so fast. ”

Li Shijie was excited for a while and then became a little lost, "I heard that they once made 1024 supercomputers, if we can get this design document, our supercomputer can run on Solaris, that will be the wings for your work." ”

"A supercomputer with 1024 processors? Lao Li, you really dare to think, how much does this cost, do we have the funds?

According to foreign quotations, it is not worth tens of millions of dollars? It is almost the same to sell our hospital, I dare not think about it, this is really unthinkable!" Song Jianjun is content and happy.

"Not now, maybe in the future, anyway, now that we have this processor, I think a lot of design can be carried out on this basis. ”

Li Shijie also knows that the grand plan of processors composed of 1024 processors is not something that can be remembered now.

However, this time, this powerful processor can be made from the Xiangjiang Dragon factory, which is a great enhancement of security for industrial applications in the mainland.

Among other things, this kind of thing, which can be produced autonomously, has no restrictions at all, and there is no need to report the usefulness and whereabouts of it to anyone. It is an extremely important supplement for applications in some special industries in the mainland.

In particular, the early SUM servers obtained through various channels in the early days, such as T1 or even older processor servers, can now be replaced by newer servers, and the speed of operation and other aspects is several times or even more than ten times the original speed.

This is like pie in the sky for these departments, and the most important thing is that the original research and development work carried out on this can be smoothly inherited.

Naturally, it is very beneficial for the next work.

Those servers and workstations that were hard to get were obtained with a lot of twists and turns, and the work done based on this platform has been accumulated for more than ten or twenty years.

It would be of great significance for these departments to be able to replace the servers of this platform with the current best-in-class performance of ultra-small computers, high-performance servers and workstations.

In view of this, the SPARC Open Alliance has led to the emergence of so many compatible server vendors.

These compatible server manufacturers are not here to make huge profits, but more to complete the modernization and transformation of servers in special key industries such as the X Academy.

Li Shijie's ability to appear at the tape-out scene is actually very revealing.

But the outside world has not had a chance to understand this.

For the vast majority of people, it's just a business.

And it's a very ordinary business.

As a computer expert, Li Shijie is a leading R&D staff at the Southwest Computing Institute, and has considerable attainments in the design and development of minicomputers.

In particular, he is familiar with the SPARC system, and according to the implementation idea of the original system design, he once led the team to imitate a 32-channel model, in which the most important motherboard replacement and IO component replacement were completed.

Now, his task is to lead the team to complete a domestic innovative R&D based on the 32-channel design, and develop the 64-channel SPARC T2 server.

Although it is only twice as much as the 32-channel SPARC minicomputer, it is quite a technical challenge, and the extra processor cores are not simply connected directly.

How to communicate with the hardware between intermediate servers and how to reasonably schedule tasks on the software system must be handled very reasonably in order to give full play to the high-performance computing power of twice the processor.

The technical complexity is completely different from that of 32.

It also involves the transformation and optimization of the Solaris operating system, thanks to Liu Ruoyi for promoting the large-scale open source action of Sun Mobile Computing, so that Li Shijie and they have the opportunity to obtain first-hand operating system kernel information, as well as the opportunity to get technical consulting services from original engineers.

These are precious to them.

Li Shijie's R&D project is listed under the name of a company called Jianchuan Computer Technology, and the funds naturally come from major scientific and technological projects.

With 64-channel super SPARC servers, Jianchuan Computer can find good market monetization opportunities in the financial industry in the mainland.

This is also Li Shijie's wishful thinking, he attracted the Miancheng Rural Commercial Bank as an application customer unit pilot, if the research and development, Miancheng Rural Commercial Bank now has the SPARC minicomputer server will use the new server of Jianchuan computer for replacement pilot.

The amount of this list signed is not too much.,It's more than 10 million in total.,It's less than 20 million.,After all, it's a compatible server.,The price requirements can't be too high.。

However, the success of this pilot will have a good demonstration effect on the development of the financial market in the future, and the low-cost super SPARC server of Jianchuan Computer will be used as a weapon, which will obviously set off a boom in the renewal and replacement of the original SPARC server field.

Anything else is not easy to say, it is the level of local city commercial banks in the financial industry, Li Shijie estimates that it is at least billions to 10 billion yuan of market space, after all, the original SUN server in the financial industry is quite popular, and there are many applications invested in it, and it is obviously the most ideal upgrade method to replace it with similar servers.

Of course, not all of these servers are super high-performance minicomputer servers, including some traditional servers such as two-socket or four-socket, and the market for super SPARC minicomputer servers alone will not be so large.

In Li Shijie's planning, Jianchuan Computer is only to do high-end, more than 8 ways to start, and other domestic friends to open the gap, so that you can use limited R & D funds to get the most powerful market weapons.

At least there are no competitors in this field of 64 Road.

32 roads should also be rare.

This is Li Shijie's magic weapon to survive in this high-end market, the price is more than one-third lower than the price of international products with similar performance, I believe that many financial companies will also think about cost performance.

The only problem is that Li Shijie needs to convince these units that his servers are safe and secure.

This is not an easy task.

Although it is convincing to speak from Li Shijie's other identity, this is the most difficult decision for Li Shijie to make.

He didn't want to add any other color to the nascent SPARC Open Alliance, hoping that it would exist in a relatively simple, impurity-free form that would reduce many barriers to technical communication.

The bigger ambition is that Li Shijie hopes to get funds through this server, and then he can go deeper into the optimization and improvement of server chips.

Wu Zuhua has the ambition to be the first in the world, and the Jiangnan Institute also has this ambition, as one of the leaders of the Southwest Institute, Li Shijie also has his own ideals.

Since SPARC was open-sourced in 2005, the Southwest Institute has invested a lot of manpower in research, and now has initially mastered the design ideas of this system and the design of existing models.

The significance of Jianchuan Computer is to obtain funds for him to develop chips, so the first step, Li Shijie has to complete the research and development of the 64-channel super SPARC server.

With this as the foundation, everything that follows can be said and have money to do, Wu Zuhua's regular army has already grabbed the position and obtained the funds for major national projects to carry out, and the Jiangnan Institute has also succeeded in getting a position from another channel.

Only Li Shijie is empty-handed, and now he is just a research and demonstration project to replace a domestic high-performance server, and the entire R&D fund is less than 10 million yuan.

He had to spend the money in two, and SPARC chip research kept a small team to follow up on it in depth and for a long time. The most important force is invested in the development of high-performance servers.

SPARC T2-based 32-socket servers and 64-socket servers are their most important projects.

No way, no matter how few people are, they have to go hand in hand to engage in two sets, the success of the 32-way server is very certain, Li Shijie has to use this to guarantee the bottom, although the 64-way prospect is better, but it may not be possible to develop it in the short term, so this is a strategic investment.

The two-pronged approach will make Li Shijie feel down-to-earth.

There are at least a few sets of 32 roads in the X courtyard that can be purchased directly, although the price is an order of magnitude lower than the original SPARC, but it is also profitable, because everyone is poor, Li Shijie also understands the other party, anyway, if the X courtyard is used well, this road can also provide a stable source of money.

This is also a very good hematopoietic channel for Jianchuan Computer.

At present, there is basically no need for any special sales channels, as long as the product is done well, and the after-sales service operation and maintenance set is in place, there will naturally be business to the door, so the uniqueness of this business is very strong.

Sales and channel expenses may be frighteningly low, but the cost of R&D and after-sales service is relatively large.

Hong kong.

The 32nm production line of the Dragon Factory, after taking over the large order production task of SPRAC T2 processor, Liang Jisong has a new business to come to the door.

Qimonda's latest generation of server DDR3 has been successfully developed, and although mass production can be done with 45nm, Clement feels that after the 45nm process technology comes out, it may not be able to achieve the lowest power consumption and heat in the market, and the best technical indicators of performance, so that it will not have an advantage in the competitive server market.

So Clement went to the Dragon Factory to talk about it.

The two companies are now strategic cooperation units, although the shareholders behind them operate in a complex offshore way, making the equity less conspicuous on the surface, directly pointing to Liu Ruoyi alone.

But a high-level executive like Clement now naturally knows who the boss behind it is.

Zhang Rujin can't refuse such a good opportunity, SPARC has a small test, after all, there are only more than 1 million pieces, and it will be completed soon, and the 32nm process is developed to make money.

Qualcomm and Broadcom's latest flagship chip design still has some time to come out, and for this new process production line, it is best to have the opportunity to debug and adapt before mass production to improve his flexible production capacity.

So Qimonda's new model of DDR3 comes at the right time.