Super Copy of Great Power Technology

Half an hour later, the test results of the first batch of products were delivered to Qiao Ruida. The good news is that the test production was not a blank sheet. Some graphene chips have passed the test and are qualified products, which means that the world's first graphene chip that can operate normally has been officially born. The bad news is that the yield rate is pitifully low, less than 0.5%, which means that most of the products are useless.

Qiao Ruida handed the test report to Lin Mengsong next to him and said, "Mr. Lin, take a look at it. The yield rate is shockingly low, but the production cost is high. With this yield rate, I would feel painful even if I produce a few test chips for my own use, let alone mass production."

Lin Mengsong took the test report, read it carefully, and said optimistically: "We can produce qualified products, which means that our process for producing graphene chips is generally fine and can work. It doesn't matter if the yield rate is lower. We just need to find the problems step by step and continuously optimize them. Sooner or later, we can increase the yield rate and meet the mass production requirements. When we were working on silicon chips before, wasn't this how we got through every time we hit a new process node?"

Lin Mengsong has extensive experience in impacting process nodes. Not to mention his time at TSMC, after coming to Rockchip, he has led the R&D team to successively overcome the 28nm, 14nm, 7nm, and 5nm process nodes, allowing Rockchip to always stand at the forefront of chip process technology, and has not wavered in the past two years.

Half a day later, all batches of graphene chips were completed and the test results were presented to everyone. Perhaps because the engineers and researchers who operated the equipment had become more proficient, the yield rate of the subsequent batches of products showed a gradual upward trend, and the yield rate of the last batch of products was close to 1%. After all five batches were completed, the overall yield rate was 0.79%, and a total of 370 qualified chips were produced. Put together, there was only a small pile that could fit in a pencil case.

Later, the performance test report of the graphite chip named "SRD-1" by Qiao Ruida was also delivered. After reading this test report, all the researchers on site, including Lin Mengsong, were excited.

"The single-core score is 3177, and the multi-core score is 12115. Isn't this a single-chip microcomputer chip? How come its performance is even better than the M1 integrated processor chip we produce?"

"It's not just about performance. Look at the power consumption, which is less than 0.5W. Look at the heat, which is 34 degrees, which is only a dozen degrees higher than the room temperature, and this is without any cooling facilities."

"I remember that in order to reduce the difficulty of production, the process of this chip is only about 100nm. In other words, the 100nm graphene chip beats the 7nm silicon chip, and the number of transistors between the two differs by hundreds of times. The performance of this graphene chip is too powerful!"

"Great! With this graphene chip, the day when our Reda Group dominates the mobile phone and PC fields is just around the corner."

"Don't be too happy yet. Even if the performance of graphene chips is amazing, it will be useless if the yield rate cannot be improved and mass production cannot be achieved."

"Yes, our focus in the future will still be on optimizing the production process and improving the yield rate. We will strive to increase the yield rate to 10% within one month and reach the mass production balance point within three months."

The production technology of graphene chips has been initially developed. Although the current yield rate is not high, regardless of the cost, it is still no problem to produce some products for internal use of Ruida Group. In the next few days, Qiao Ruida did not go to the laboratory, but stayed at home. He found a graphene processor used in Apple notebooks and a multi-core graphene processor used in Apple servers on the Internet. After analyzing it with the gold finger, he reproduced the design drawings. At the same time, in order to support the Kaitian system and the existing hardware protocols, some modules of these two processors were modified.

A week later, Qiao Ruida handed Lin Mengsong an encrypted USB flash drive and said, "These are the two graphene processors I designed. You can take them for tape-out testing to see how they perform and whether there are any bugs."

"Okay, Mr. Qiao, how many nm process will be used for this tape-out test?" Lin Mengsong took the USB drive, nodded, then remembered something and asked.

"How much of the NM production process have you mastered now? Let me make it clear first, the ones with a yield rate of less than 1% will not be counted?" Qiao Ruida asked back with interest. He also wanted to know how much progress the research team had made during the week when he was away.

"Mr. Qiao, I think the 55nm process is the best choice, taking into account both performance and yield rate. This is also the first process node that we should overcome in our graphene chip production and testing line." Lin Mengsong thought for a long time before answering seriously.

"Well, let's choose the 55nm process. As long as the yield rate can be improved, relying on the 55nm graphene chip, it will not be a problem to beat all existing silicon chips." In this world, only Qiao Ruida knows how powerful graphene chips are. It is not an exaggeration to say that they crush silicon chips. Graphene chips have many advantages, high performance, low power consumption, and low heat generation, which are just the basics. Graphene chips have a very wide temperature adaptability range and can work in extremely harsh environments. There is no problem with being directly exposed to space. When quantum computer technology matures, graphene chips can also seamlessly connect to quantum computing technology. With a slight adjustment to the production line, quantum chips can be produced. It can be said that a one-time investment will benefit in the long run.

One day later, the first batch of graphene PC processors rolled off the production line, and the yield rate was still very low, just over 3%. However, this yield rate has made Qiao Ruida quite satisfied. In just one week, the yield rate can be increased by more than three times, which shows the strength of Lin Mengsong and his R&D team. Moreover, the area of ​​this PC processor is several times larger than that of the last single-chip microcomputer chip, and it is also many times more complicated. The production process has also been increased from 100nm to 55nm, and the production difficulty has increased by more than ten times. Under such circumstances, it is still very difficult to maintain a pass rate of more than 3%.

"Mr. Qiao, these qualified products have not yet been silk-screened. What do you think should be named this first graphene PC chip?" Lin Mengsong asked, playing with a stamp-sized chip in his hand.

"This chip uses a similar architecture and instruction set to the M1 chip, but its performance is much improved, so let's call it the M2 chip." Qiao Ruida answered very casually, not taking the chip naming too seriously at all.

"M2? This name is really perfunctory. Well, the name is just a code name. As long as the performance is strong enough, there is no need to worry about sales. Xiao Wu, come here and use the laser engraving machine to print the M2 on these chips." Lin Mengsong waved to a researcher and gave him the task of silk-screening qualified chips. After getting these hundreds of qualified M2 processors, Qiao Ruida personally designed and made motherboards, soldered processors, various components, built test platforms, installed systems, and ran various tests. A few hours later, the performance test results of the M2 processor were freshly released. All researchers who participated in the chip testing were very excited. The reason was that the actual performance of this M2 processor was really too powerful. From the running score point of view, the performance of the M2 processor is more than fifty times that of the M1 processor. In terms of graphics cards, the running score of the M2 processor's integrated core graphics card is more than twice that of NVIDIA's newly released flagship graphics card Geforce GTX 760. It is no longer a dream for integrated graphics to beat independent graphics cards, but a fact in front of us.

The subsequent game tests and video transcoding tests also proved that the M2 processor is powerful and not just a victory in running points. The M2 processor can play almost all 3A masterpieces. Even with full special effects and resolution adjusted to 4K, it can maintain more than 100 frames and run very smoothly. In contrast, the Core i9 + GeForce GTX 760 combination can only maintain more than frames when adjusted to the same resolution. Once it enters the battle scene, it starts to drop frames and freeze.

Of course, the M2 processor and this temporary test platform are not without shortcomings. Due to the imperfect driver, some games cannot be opened, and some games that can be run will also have some texture errors and inexplicable freezes, affecting normal games. This requires the Reda Research Institute to organize software and hardware engineers to continuously optimize this system.

In addition, the M2 processor is in a leading position, so fast that other hardware on the platform, such as the fourth-generation memory, the first-generation M.2 interface SSD hard drive, etc., cannot keep up with the processing speed of the M2 processor, forming a bottleneck, which, to a certain extent, affects the performance of the M2 processor. It is difficult to optimize this point, unless one day the Ruida Group is strong enough to independently formulate a set of its own interface standards and promote them so that many third-party manufacturers can accept this standard.

What makes Qiao Ruida even more delighted is that such a powerful M2 chip has very good power consumption control. The standby power consumption is 2W, and the double-baking power consumption will not exceed 5W. The power consumption of the whole machine is slightly higher, with a standby power consumption of 5W and a double-baking power consumption of no more than 10W, which is a model of energy saving. And thanks to the super conductive properties of graphene, the heat generated by the M2 chip is very low. You only need to stick a thin piece of aluminum sheet to solve the heat dissipation problem, and there is no need for fan heat pipes or the like. This means that laptops and host products that use M2 as processors can completely adopt passive cooling systems, and the annoying fan noise will be completely gone.

Later that day, the server processor designed by Qiao Ruida also rolled off the production line. This processor uses the same core and architecture as the M2 processor, but the number of cores has increased to 72, including 4 small cores, 20 medium cores, and 48 large cores. In order to further improve the multi-threaded processing capabilities of the processor, Qiao Ruida added hyperthreading design to the medium and large cores, so this 72-core processor has up to 144 threads, and its performance is amazing.

Subsequent performance tests have also proven this point. This new graphene processor is more than fifty times better than Intel and AMD's top server CPUs in terms of performance, but its power consumption is less than one-tenth of the latter. It is conceivable that once such a graphene processor product solves the yield problem and is launched on the market, it will set off a storm in the server field.

That afternoon, Qiao Ruida called Luo Yuankai, Lin Mengsong, and several other senior executives of Ruida Wafer Factory to the conference room for a short meeting.

"Before the formal meeting, you should take a look at these two documents first." After that, Qiao Ruida signaled Lin Mengsong to distribute a copy of the performance test report of the two graphene chips to each executive in the conference room.

"Oh, the graphene chip has produced research and development results so quickly, Mr. Lin, you are really quick!"

"I don't dare to take credit for this. It's mainly because of Mr. Qiao's guidance after he joined us that we were able to master the production technology of graphene chips in just one month. Otherwise, we old guys wouldn't be able to do it even if we had another six months."

"Even if the graphene chip is successfully developed, it is nothing more than adding two more production lines. Is it worth gathering so many people to hold a meeting to study it?"

"MY, GOD, someone pinch me to make sure I'm not dreaming. The performance of graphene chips is dozens of times that of silicon chips. How is this possible!"

"Are you sure today isn't April Fools' Day? You can't play a joke like this."

"If this report is true, the silicon chip, which has flourished for decades, will soon face its doom. At least in the field of high-end chips, graphene chips will soon replace silicon chips."

"Fortunately, fortunately, the graphene chip with such amazing performance was born in the laboratory of our Rockchip wafer factory. This way, we have the upper hand and can attack or defend, and we will always be invincible."

"So, are we still going to go ahead with our earlier plan to build ten new 5nm silicon chip production lines?"

"With the graphene chip production technology, of course the construction of graphene chip production lines will be given priority. As for the 5nm silicon chip production line, it will soon become an obsolete technology. Why build so many?"

When they first received the two test reports, all the company executives, including Luo Yuankai, were somewhat unimpressed. In their opinion, even if graphene chips were successfully developed, their performance would not be much better than silicon chips, and their production costs would be much higher than silicon chips, and their technology was not mature. It would take at least 10 to 20 years for graphene chips to completely replace silicon chips, so there was no need to rush to develop them now.

But after reading the details of the two reports, everyone was shocked. The performance of graphene chips is dozens of times better than that of traditional expensive chips. This is not just a simple technological iteration, but a reshuffle of the entire chip industry. (End of this chapter)