109th line, I will go

121.1 trillion!

The second phase of the "Cambrian 2004" project research, understatement, has crossed the threshold of one trillion times.

In the June 2004 issue of the global supercomputer Top_500, Earth-Simulator, which occupies the top of the list, Linpack calculated 35.86 trillion times. The second phase of the Cambrian 2004 project Cambrian 2004B is astonishingly three times that of Earth-Simulator. It not only rushed to the highest position in the world at one time, but also set the record to 121.1 trillion times.

Cambrian 2004B is China's first trillion-level supercomputer and the world's first trillion-level supercomputer.

Number one in the world and number five in the world are completely different concepts. At least when the future scientific research news reports, Cambrian 2004A is the international advanced level, and Cambrian 2004B must be the world's leading level. If the Cambrian 2004 project can make persistent efforts to complete the third phase of the study, and develop a Linpack calculation value of 300 trillion supercomputers, and monopolize the top_500 supercomputer list for a certain period of time, the Cambrian 2004 project and its follow-up will definitely be able to become the representative of the 20th China. The benchmark for the new image of the first century.

The fact is also true.

It was verified that the Cambrian 2004B was a truly one-hundred-terabyte supercomputer, and the national level of attention quickly increased several times.

In fact, the fifth in the world is also rare, Cambrian 2004A was included in the national sight. However, whether heterogeneous supercomputers are universal, what role Smart Universe Information Industry Corporation can play in the field of high-performance computing, etc. are still under discussion, especially the wrangling between the interests of various parties, and it is very clear in a short time Positive results. For example, Lenovo and Suguang accuse the smart information industry company of only using money to dig people. The Xiao Lei team who really made achievements are all from Suguang and Lenovo.

The country is a game of chess. It is necessary to consider the interests of Sugon and Lenovo, as well as the interests of other research units.

Seeing such difficulties, Xiao Lei firmly gritted his teeth, encouraged team members and recruited new blood as appropriate, and desperately accelerated the second phase of the Cambrian 2004 project. Xiao Lei is already familiar with the basic principles of heterogeneous supercomputers and is confident to shorten the research time. Because Cambrian 2004A has verified the theory of Cambrian 2004B, and the theory allows it to be enlarged, Cambrian 2004A can be transformed into Cambrian 2004B. In other words, the Cambrian 2004A was a perfect success, and the Cambrian 2004B had a 98% chance of success. Xiao Lei doubled the second-stage research goal, and the chance of success fell to 95% at most.

The first stage is verification, the second stage is realization, and the third stage is to further improve heterogeneous supercomputing research. From the second stage to the third stage, the heterogeneous theory will be practiced and supplemented, and it will take a long time to study some difficult problems that are temporarily inconclusive.

Therefore, due to adequate preparations, the Xiao Lei team produced the Cambrian 2004B with 121.1 trillion Linpack calculations in less than three months.

The entire scientific research community in China was taken aback: You can't use satellites like this.

On May 24, 2004, Cambrian 2004A made its debut with 11.6 trillion Linpack calculations.

On August 15, 2004, Cambrian 2004B announced the arrival of the Demon King with 121.1 trillion Linpack calculations.

In just three months, computing performance has skyrocketed tenfold. When will supercomputers be as simple as blowing a balloon?

However, it is true.

Xiao Lei's team had enough confidence in Cambrian 2004B, and Wei Dongsheng ordered them to open their minds and accept appraisal from relevant departments. No fraud or exaggeration, Cambrian 2004B is indeed a tens of billions of supercomputers.

The dawn was silent.

A few months ago, facing the Cambrian 2004A, Shuguang was not convinced.

Among the global supercomputer Top_500 in June 2004, Dawning 4000A ranked 11th and Cambrian ranked 5th. However, Sugon 4000A developed by Sugon has a Linpack calculation value of 8.061 trillion times and a peak value of 11.264 trillion times. Simply comparing the performance values ​​of two supercomputers, the difference is obviously very limited. Sugon believes that as long as the country is willing to fund support, it will be able to develop a supercomputer that will fully surpass Cambrian 2004A in the coming year. The so-called heterogeneous supercomputer, if the theory is really feasible, handed over to Sugon, in order to maximize utilization.

If Xiao Lei questioned whether Sugon could develop a Cambrian 2004A-level supercomputer at the end of May, Sugon would definitely answer with confidence: "As long as we have sufficient funds, we are confident that we can build a supercomputer with 12 trillion Linpack calculations."

Now, Xiao Lei asks again: "Can Sugon develop a supercomputer with a hundred billion Linpack calculations?"

Dawn has been unable to answer confidently.

Facing the Cambrian 2004B, Suguang could only be awkwardly silent.

Because, don't say that a supercomputer with a trillion Linpack calculations is a supercomputer with a peak value of one trillion, and it is far beyond the capabilities of Sugon.

In May 2004, the Huaguo Meteorological Administration invited companies such as Lenovo, Sugon, Yinhe, Cray, HP, IBM, NEC, and SGI to participate in a tender for high-performance computer clusters. The China Meteorological Administration's procurement is a capped bidding. The purchase amount is fixed at 20 million U.S. dollars, that is, within the purchase amount of 20 million U.S. dollars, whoever can provide the cheapest and most beautiful high-performance computer cluster system wins the bid. At the bidding meeting, IBM's bid was a peak of 20 trillion times.

Facing IBM's peak 20 trillion bidding scheme, Sugon was horrified and helpless, and had to withdraw sadly.

Look, Sugon has no confidence to develop a high-performance computer cluster of 20 trillion times.

A supercomputer of the Cambrian 2004B level, Sugon can no longer imagine.

In fact, it is more than dawn.

At that time, the Smart Universe Information Industry Company took a special channel and also participated in the bidding meeting of the China Meteorological Administration. Faced with IBM's confident smile, conscious that everything is under control, the Wisdom Universe delegation calmly shouted: "50 trillion times."

The Wisdom Universe delegation stated that the Wisdom Universe information industry can provide the China Meteorological Administration with a peak of 50 trillion high-performance computer clusters.

The tender meeting was in an uproar.

Not only IBM mocks the Wisdom Universe, but the Meteorological Bureau suspects that the Wisdom Universe is making trouble.

In the global supercomputer Top_500 in the June 2004 issue, Earth-Simulator, which occupies the top of the list, has a calculation value of 35.86 trillion times by Linpack and a peak value of 40.96 trillion times. The world's No. 1 Earth-Simulator has a peak value of only 40 trillion times. You don't know which rock gap has emerged from the intelligent universe information industry, which is qualified to make a peak price of 50 trillion times.

During the bidding process, Xiao Lei’s team successfully developed Cambrian 2004A, Linpack calculated 11.6 trillion times and the peak was 33.5 trillion times. Although the peak value of 33.5 trillion times is still some distance from the peak value of 50 trillion times, this result is already very good. Looking at the global supercomputer Top_500 in June 2004, there is only Earth-Simulator with a peak of 40 trillion and only Cambrian 2004A with a peak of 30 trillion. The mainstream of the era is more than ten trillion. To more than twenty trillion times.

The problem is, someone has to believe it.

Look at Earth-Simulator, Linpack calculates 35.86 trillion times, the peak is 40.96 trillion times; Look at Dawning 4000A, Linpack calculates 8.061 trillion times, the peak is 11.264 trillion times, the gap between Linpack calculation and the peak is limited. . In Cambrian 2004A, Linpack calculated 11.6 trillion times and the peak was 33.5 trillion times. The difference between the two values ​​is also far away. How unsuitable is Linpack calculation? The so-called heterogeneous supercomputer is really not a scientific research gimmick?

Although Cambrian 2004A did not falsify and was not afraid of verification, public opinion feedback, from correct verification to doubtful, maybe this to true, it takes time to slowly change their mentality.

What do you think, in just over two months, the high-performance computing community has not yet reacted to the Cambrian 2004A shock. The Smart Universe has surprisingly developed a Linpack calculation value of 121.1 trillion times and a peak value of 345.1 trillion times. Wu Ji 2004B. The unspoken rules of the industry pay more attention to propaganda peaks. If the Cambrian 2004B really reaches the peak of 345.1 trillion times, it can already be called a three trillion times supercomputer.

The technological background of the Wisdom Universe has reached its peak more than 300 trillion times.

If everything is true, heterogeneous supercomputers are true, and Cambrian 2004B is also true. It is easy for the Smart Universe Information Industry to supply a high-performance computer with a peak value of 50 trillion operations to the China Meteorological Bureau.

The peak value of 20 trillion times is already in 2004, and the peak of 345.1 trillion times Cambrian 2004B is simply desperate. You can't even make a 20 trillion supercomputer, but can you make the most of it by handing it over to you?

Dawn’s contribution cannot be slandered.

If Wei Dongsheng did not return with 30 years of memory, Dawn must be an important part of the country's development of high-performance computing. Sugon cannot develop a high-performance computer cluster with a peak value of 20 trillion operations. With the development of technology and the hard work of scientific researchers year after year, they will surely be able to make a super high performance of 2,000 trillion operations. computer.

However, the future is the future, and the present is the present.

Now, the Wisdom Universe is stronger than the dawn, and strong enough to crush the country of China by oneself.

No, it is your own power that crushes the world.

The blue giant IBM had to kneel when it hit the wisdom universe.

Time waits for no one. Wei Dongsheng respects Suguang's contribution, but will not passively wait for Suguang to grow.

At this time, China is my best.

proceed if you can?

If I do, I will go.

Wisdom Universe has the capital and strength, so it should take responsibility and lead the early rise of China's high-performance computing industry.

When the Cambrian 2004B looked down on the mountains and looked down on the heroes, the interest disputes with Shuguang, Lenovo and other companies came to an instant result.

the reason is simple.

The state's support for the high-performance computing industry is based on the thinking and measurement of the interests of the 1.3 billion people across the country, and based on the thinking and measurement of national homeland security. The state's support for high-performance computing is not aimed at supporting companies such as Sugon and Lenovo, and it is even more impossible to take care of Sugon and Lenovo unconditionally as a nanny. The interests of mere units and departments cannot go against the strategic needs of the country in a game of chess.

If there is no Wisdom Universe without Wei Dongsheng, the Ministry of Science and Technology and other national institutions will definitely support Sugon and Lenovo, and provide funding policies and other support; but now there are better choices, why not support a more promising and powerful Wisdom Universe What?

If you like to re-take the road of the future, please collect it: (lightnovelpub.net) Re-take the road of the future. The literature is updated fastest.