1st of May.
A piece of great news came from the Supercomputing Research Center in Tianjin City.
After nearly a year, Xing Hai Technologies' first self-developed supercomputer, the Supernova 1, was successfully developed.
Since then, Xing Hai Technologies had a self-developed supercomputer system. They could save a lot of money on the purchase of a supercomputer.
Regarding the performance of the Supernova 1, the project leader, Zhao Zhang Lai, explained to Chen Jin in a remote video conference.
Zhao Zhang Lai was in his forties. He wore glasses and a white lab coat. He used to be an associate professor at the University of National Defense. He had participated in the development of the "Heavenly River 2" supercomputer. In this project, Zhao Zhang Lai was the "secondary architect." His role was very important.
In order to poach him, Chen Jin not only offered him an annual salary of 10 million, Xing Hai Technologies also signed a "Strict Confidentiality Agreement" with the University of National Defense. They only agreed to let him go after Xing Hai Technologies paid 100 million in technology licensing fees.
But in reality, the mechanism used in the Supernova 1 was brought over by Chen Jin from the Base of Operations. It was obviously more advanced than the architecture of the "Heavenly River 2." He poached Zhao Zhang Lai only because he was more familiar with this area. No matter how advanced a technology was, it had to be implemented by experienced people.
As for the University of National Defense Science and Technology, they had already developed the "Heavenly River 3E" with a computing speed of ten billionths of a second. The "Heavenly River II" technology, which was obviously outdated, was no longer as important. Therefore, it was not a big deal that some of their talents had been poached.
However, the 10 million annual salary offered by Xing Hai Technologies caused a sensation within the University of National Defense. Many leaders were furious. "This is simply disrupting the military's morale!"
…
Zhao Zhang Lai introduced.
"President Chen, the Supernova architecture that you provided us is at least two generations more advanced than the 'Heavenly River 2.' It is on par with the 'Heavenly River 3E.' Theoretically, we can develop a 10 billion pegasus class supercomputer based on this architecture.
"However, you said that the company was short of supercomputers and needed them urgently. So, we started small and built a small 'Supernova 1'."
"The peak computing speed of the Supernova 1 is about 1.25 billion pegasus per second. Its continuous performance is 1 billion pegasus per second.
"The overall standard is comparable to the Light of Taihu Lake before it was upgraded. Its performance is ranked third in the country."
"But due to the advanced architecture, the power consumption ratio of the Supernova 1 is very outstanding. We achieved 9G/W, which exceeds the Light of Taihu Lake A's 6.5G/W, and is comparable to the Heavenly River 3E's power consumption ratio … It's one of the best in the world, almost reaching the limits of traditional computers."
"With this power consumption ratio, we can save hundreds of millions of dollars in electricity bills every year."
Chen Jin nodded. The most expensive thing about a supercomputer was not the construction costs of the hardware, but the high electricity costs!
For example, the "Heavenly River 2" was a big power consumer, with a power consumption of 17,808 kilowatts. After installing a water-cooling system, the power consumption would reach 24 megawatts … equivalent to the power provided by a small nuclear reactor.
The annual electricity bill was as high as 100 million RMB.
The continuous computing performance of Supernova 1 was three times that of Tianhe-2. However, Supernova 1 also had to save 30% of its energy. The annual electricity cost was about 66 million yuan.
It was even more economical to use.
"In addition, the supercomputer architecture that President Chen provided to me has unimaginable stability. The failure rate of a normal supercomputer is quite high. Every time a computing core breaks down, it takes half an hour to repair it. There are tens of thousands or even hundreds of thousands of computing cores in a supercomputer. Often, one or two of them will break down … Such a supercomputer cannot meet the needs of long-term use. It can only run some short-term computing projects.
Many of our company's programs and applications require long-term operation and calculations … We cannot allow the supercomputer to break down too often.
"Our Supernova architecture uses self-detection and intelligent parallel technology. One or two computing cores breaking down will not affect the normal operation. It can be directly replaced when the supercomputer is switched on … Even if half of the cores break down, the supercomputer can still be used normally.
"In terms of stability, the Supernova architecture is not weaker than the architecture used in the Heavenly River 3E.
Moreover, the Supernova architecture has great expansibility. We can completely develop an E-grade supercomputer on the basis of this architecture. "
"…"
Zhao Zhanglai said. He could not finish talking about the various advantages of the Supernova architecture. In short, his decision to jump ship to Xing Hai Technologies was the right one. Not only did the research on supercomputers here have a sufficient budget, but the future was also bright.
He felt very satisfied.
Chen Jin nodded repeatedly.
In many people's impression of supercomputers, it was as simple as stacking a large number of CPUs together. It was shortened to "stacking cores."
In reality, it was far from that simple. To make tens of thousands or even hundreds of thousands of computing cores work together and process one thing at the same time, the massive amount of data generated could easily exceed the transmission bandwidth of communication lines, exceed the read and write capacity of memory chips and hard disks … These bottlenecks could only be overcome by advanced architectures.
In the field of supercomputing, there was a reason why the Z nation's supercomputers had repeatedly dominated the rankings and were often ranked number one in the world. It was because their chips were not good enough. Since the 1980s, or even earlier, the Z nation's researchers could only develop parallel CPU computing horizontally, compensating for quality with quantity. In that era where everything was backward, the older generation of researchers left behind the younger generation in the field of computing.
This was also the reason why the Z nation's researchers could still create the world's number one supercomputer by using domestic CPUs that were several generations behind, even after foreign countries banned the sale of advanced CPUs.
The CPU, memory chips, hard disks, and other core components used in the self-developed "Supernova 1" by Xing Hai Technologies were all domestic. In order to reduce the purchase price, Xing Hai Technologies even issued a huge order that was enough to produce 15 "Supernova 1" units in one go. This shocked the manufacturers.
Therefore, the cost of the "Supernova 1" was successfully reduced. Excluding the research and development expenses, the cost of the hardware alone was about 500 million dollars.
"Indeed, it's better to build than to buy," Chen Jin lamented.
He said to Zhao Zhanglai, "Assemble five Supernova 1 units for me first. The company needs them! After this task is completed, immediately dive into the development of the Ai class supercomputer and develop the 'Supernova 2' with a computing capacity of 10 billion Flops. "
"Understood, President Chen."
Zhao Zhanglai nodded.
…
Another few days passed.
News that was not too optimistic came.
Ma Xiaobing, the person in charge of the Star Cloud System project, said, "Mr. Chen, the number of programmers using the Star Cloud Studio to try to develop the Star Cloud APP has dropped from a peak of 1 million to less than 100,000. It's a steep decline."
It's a precipitous drop. This is because Google has forcefully pushed out an 'anti-swipe patch.' As long as the phone is swiped to the Star Cloud system, the underlying hardware of the phone will be damaged and become unusable. Currently, 90% of the Android phones in the country have this patch. This is simply a rogue push! A virus push! The purpose is to block our Star Cloud system! "
This also directly confirmed the rumor that there were backdoors in foreign software or operating systems. It was indeed true.
Backdoors were not only in the software, but also in the underlying hardware. Otherwise, how could a change at the software level damage the hardware?
Therefore, after the "backdoor incident" was confirmed, on one hand, the Z nation's Information Security Department was investigating and would soon issue a fine to Google.
On the other hand, the confidence of the vast number of programmers who wanted to participate in the development of the Star Cloud application was greatly shaken. One after another, they chose to give up.
Because things had developed to this point, it was no longer just a software problem. No matter how good the Star Cloud system was, it was useless. As long as the backdoor in the underlying hardware was still there, the Star Cloud system could still collapse.
Unless Xing Hai Technologies had a way to achieve complete autonomy in both hardware and software, the Star Cloud system could not be built.
A senior programmer sighed, "The death of the Star Cloud system will give us a wake-up call. Unless we achieve complete autonomy in both hardware and software, the backdoor will always exist."
But was the Star Cloud system really going to die?
You've already exceeded your reading limit for today. If you want to read more, please log in.
Login
Select text and click 'Report' to let us know about any bad translation.
