1991 Starting from the core - Chapter 994

c992 finfet

“Brother, you can’t be in such a hurry.” Su Yuanshan smiled and made Qin Weimin a cup of tea, then sat down next to him.

Not long after he received a call from Liang Mengsong in the car, Qin Weimin called him. After learning that Su Yuanshan was on his way back, Qin Weimin took the time to come to the office and wait for him.

Qin Weimin has always been known for his calmness, but it is rare to be so anxious.

“It’s not that urgent. It’s just that I haven’t seen Wenjie for a long time, so I have to go and take a look, right?” Qin Weimin smiled: “As for the other things, that’s just a matter of convenience.”

“Hahaha! If you want to see Brother Tang, it’s not just a phone call away? How dare he not come over?” Su Yuanshan laughed immediately, exposing Qin Weimin’s little lie.

But then he nodded and said: “You can take a look in the past. In fact, I don’t know the process of their low-power version now, and I also agree that performance and power consumption need to be synchronized.”

——When a manufacturing process is first launched, it must require a large number of orders to drive down the cost. In the pure foundry model of Deyuan, in the past, due to the unreserved support of Yuanxin and Xinghai, two Fabless companies pursuing advanced processes, the maturity period of its process was shortened to a level that others could not match. This has resulted in the current situation in which Deyuan is the dominant company and the strong are always strong, seizing most of the interests in the OEM field.

Historically, Intel relied on its advantages in process technology and its large orders for FPGAs and processors to lead the process technology all the way to the point where it was the first to launch the 22nm finfet node, which was even two years earlier than TSMC. .

However, as TSMC broke through the 22nm process, especially the 22nm low-power process, a large number of orders allowed TSMC to quickly achieve breakthroughs in the process nodes, and eventually overtook Intel. During this period, the low-power process actually played a very critical role. Role – to be more specific, the advent of the smartphone era. Intel, on the other hand, has begun a long journey of squeezing out toothpaste.

Since Deyuan has always been responsible for Yuanxin and Xinghai, and later added AMD, the most advanced design solutions of these three desktop processor companies, every process node of Deyuan prioritizes high performance, and then can take care of low-end products. Power consumption – this is one of the main reasons why Qin Weimin is anxious this time.

After all, the development direction of YX architecture is low power consumption.

Hearing that Su Yuanshan recognized his appeal, Qin Weimin was completely relieved, smiled and slapped Su Yuanshan on the shoulder: “Then it’s settled, I’ll go with you tomorrow. We really can’t delay, RISC-V is progressing very quickly. I am really worried about what Silicon Valley and Europe will do if it takes too long – oh, and Bangalore too. Last year, the third brother was very ambitious and said that Fully support RISC-V and make it a national instruction set…”

“Well, although I’m not very optimistic about the third brother’s government, the RISC-V Foundation really shouldn’t be underestimated.” Speaking of which, Su Yuanshan also sighed, feeling the same as Qin Weimin.

As time goes by, the impact brought by the rise of Yuanxin is no longer at the level of butterfly wings, but at the level of banana fans. It can even be said that today, in the field of semiconductors, the comparison of influence between the East and the West has completely changed. .

——In the field of chip design, although the market share of Zhiyuan EDA has “dropped” to just over 50%, this is entirely because with the deepening of the information technology revolution, a large number of small chips have emerged. Design companies, as well as several traditional EDA companies in the West, have “grabbed” the past market through continuous mergers and development and continuous subdivision of functions. In essence, Zhiyuan EDA is still the most complete and complete EDA product at present, and there is no other EDA product. one.

In the field of semiconductor equipment and foundry, the rise of Acer and Deyuan is no longer a reversal of status, but has directly caused the “roots” of semiconductors to fall in the East.

Not to mention the YX architecture. Since defeating ARM, the YX architecture has continued to evolve with mobile phones and has become the most influential architecture in the world.

Faced with this situation, Silicon Valley is obviously unwilling to give up. Among so many fields, if you want to reshuffle the cards, the only field that is convenient to start with is actually low-power architecture.

Therefore, RISC-V came into being. As soon as it was born, it received strong support from various universities in the West, as well as established companies such as IBM, Intel, Motorola, and Qualcomm. It has developed extremely rapidly in the past two years, not to mention a large number of outstanding IPs, and it also has an open source spirit… …No matter how you look at it, it seems that it is trying to bring down RISC-YX.

How do you say that?

“It’s important to me without you.” – This is the collective thought in the Western semiconductor field at this time.

Therefore, according to the latest news, it is very likely that Qualcomm will adopt the RISC-V architecture in 4G chips.

Therefore, Su Yuanshan can completely understand Qin Weimin’s anxiety – Deyuan is a professional foundry, and its business is independent and fair. It is impossible to say that the RISC-V architecture comes to the door and turn it away.

Therefore, if Deyuan can launch low-power technology earlier and let the IP of risc-YX architecture mature earlier, it will obviously help maintain the status of YX architecture.

*

*

The next day, Qin Weimin and Su Yuanshan set off for Shanghai.

After leaving the airport, the two went straight to Deyuan.

“Brother Tang, you didn’t come to pick us up at the airport?”

At the gate of Deyuan, Su Yuanshan looked at Tang Wenjie standing at the door like a hill. He stepped forward and punched him in the stomach. As a result, he found that he had punched a wall…

“Haha, I’m jet-lagged.” Tang Wenjie laughed and gave Su Yuanshan a bear hug. Then he took Qin Weimin’s hand and held his shoulder: “Senior brother, long time no see.”

“Well, long time no see – you came here from Singapore and it’s a hell of a time difference.” Qin Weimin looked at Tang Wenjie with a smile and rubbed his arm without leaving any trace.

“Yes, you still pretend to be a master – where is Mr. Liang?” Su Yuanshan looked through Tang Wenjie’s back towards the park, and saw that except for a few employees in the park, Liang Mengsong was not found. Hmm… He is relatively satisfied.

“The office is waiting for you, we won’t engage in formalities.”

“Well, no, let’s go.” Su Yuanshan smiled slightly, convinced.

The three brothers then got on the open-top sightseeing car in the park and headed straight to the technology center, laughing and talking.

When they arrived at the technology center building, Liang Mengsong brought a group of people waiting there, but everyone was acquaintances, so they went straight to the conference room without much greeting.

…

“Because the samples have just been released, it’s hard to talk about the yield rate.” Liang Mengsong looked a little thinner than when he last met at the Science and Technology Committee meeting. He must have worked a lot of overtime recently, but he was extremely energetic and enthusiastic. Di pointed at the LCD panel with a pen and said: “But the process is no problem, and we can continue along the way.”

“The process is fine, does that mean the PDK can be updated?”

“Well, it’s actually being updated. I estimate that EDA will follow up within a month at most.”

Su Yuanshan breathed out lightly, and then he glanced at Qin Weimin again: “This is the high-performance craft that goes first, right?”

“Even if we go to lower power consumption first, we still need someone to agree…” Liang Mengsong said with a rare joke: “Lao Gao comes to my place eight times a day…”

Everyone laughed together.

Gao Xiaodi is the first to get to the moon near the water and the tower, so he must have been running here a lot.

“What about the time for the low-power version?” Su Yuanshan continued to ask for Qin Weimin.

“We can only say as soon as possible. If we have to give a time… three to six months.” Liang Mengsong gave a fairly loose time.

Su Yuanshan and Qin Weimin looked at each other and nodded at the same time.

——Although this time is a bit wide, the maximum period of six months obviously means that Deyuan is planning to work overtime.

Su Yuanshan silently counted the time and found that the finfet process would officially enter the mass production stage by the end of next year at the latest. Then he compared it with his memory… and he was delighted to find that Moore’s Law was advanced by Deyuan this time.

——In memory, it seems that the first finfet chip was launched by Intel in 2011.

This discovery made Su Yuanshan take a long breath and was quite emotional.

——If Intel still does not change the IDM model, then in the next few years, toothpaste squeeze will probably be a foregone conclusion. At the very least, in terms of process nodes, there is a high probability that it will continue +++.

if that is the case……

Not to mention AMD and Cyrix, maybe even FarCore can pull down Intel by relying on transistor density.

——You know, Jim Keller has fully taken over the architectural design of Yuanxin’s next-generation processor. Although you can’t overly flatter a certain person, he is really awesome!

*

*

Now that Su Yuanshan and Qin Weimin are in Shanghai, there is no shortage of gatherings. Especially Qin Weimin. Since the separation of the YX structure, Qin Weimin has also left the Yuanxin Science and Technology Committee, so that he has very few opportunities to meet with Li Mingliu, Gao Xiaodi and others. Today is the first meeting this year.

Therefore, that night, Gao Xiaodi clamored to kill Qin Weimin no matter what.

But as we all know… Mr. Qin’s drinking capacity is not comparable to that of ordinary people – as Su Yuanshan’s senior brother and Academician Su Xinghe’s chief disciple, Qin Weimin is the kind of person who can stop drinking for his teacher. The ordinary people can’t understand his depth at all.

Not to mention that there are three brothers gathering together tonight.

So, Gao Xiaodi screamed and simply got himself down, and by the way, Li Mingliu was also dragged down.

…

After sending away all the local landlords, Su Yuanshan, Qin Weimin, and Tang Wenjie returned to Su Yuanshan’s room.

“Li Mingliu and Gao Xiaodi can’t do it…” Tang Wenjie sat on the largest sofa in a majestic manner, and happily took the teacup handed over by Su Yuanshan: “I thought they were so powerful, but in the end I didn’t do anything. After exerting force, they both lay down.”

“Haha, that’s why senior brother is awesome.” Su Yuanshan gave a thumbs up to Qin Weimin, who was blushing: “To be honest, I really admire senior brother… Let me interview you, how do you maintain a balance between drinking and being sober? ”

“Born.” Qin Weimin took a sip of tea and said indifferently: “It must be because of the strong digestive enzymes.”

“Uh… where’s the liver?”

“There is no problem with the liver. I did a comprehensive check-up after Wang Rui’s accident, and now I also focus on the annual physical check-up.” Qin Weimin said with a smile: “I can only say that he is gifted.”

Tang Wenjie looked envious when he heard this: “It would be a shame if you don’t become an official.”

…

After laughing, the three people naturally brought the topic back to the finfet structure this time, or in other words, further.

For so many years, Tang Wenjie has been walking on two legs.

The first one is to cooperate with Deyuan to commercialize the finfet structure and plan the subsequent improvement plan. The second one is to try new designs to cope with future limits.

Or to be more precise, it is the limit of leakage.

When the number of transistors reaches a certain level, the transistor density is not as “decisively” important as before – in a sense, the improvement and optimization of the overall processor architecture are enough to erase these advantages.

What’s more important is the leakage rate.

Just as the human heart is insatiable, technological progress is also insatiable.

“To be honest, twenty years ago, I really didn’t expect that I would touch the quantum realm in this life.” Tang Wenjie smiled and shook his head with emotion: “Now I actually looked up and found that quantum tunneling has become a barrier in front of me. The biggest obstacle…I am also a little drunk.”

“Well, the main reason is that the semiconductor industry has developed too fast in the past two decades.” Su Yuanshan took a sip of tea and said with a smile: “Silicon-based chips are coming to an end soon.”

As soon as he said this, Qin Weimin and Tang Wenjie immediately looked at him at the same time.

“Really? Have you developed some black technology?”

Su Yuanshan coughed immediately. He knew that his casual emotion had misunderstood the two senior brothers, so he immediately explained: “No, no, I just said that casually.”

“I was shocked. I thought it was about to be converted to a carbon-based chip.”

“It’s still early. The current progress of carbon-based chips…how should I put it, it’s still in the fairy child era.” Su Yuanshan squinted his eyes, thinking of the current progress of graphene.

As the discoverer of graphene, he certainly will not miss any progress in graphene. But he also knows that graphene… at least for a long time, will belong to the kind of future that “looks good”.

At present, whether it is universities or scientific research institutions, the research on graphene… apart from being used to write papers, is actually more of a “backup”.

After all, the upper limit of silicon-based chips is visible to the naked eye.

Carbon-based chips, due to their naturally excellent electron mobility, can surpass silicon-based chips in terms of greater flexibility in the manufacturing process.

The so-called “backup” is to open a door and leave it there, whether it is in the future if the silicon-based chips have no way out and there are no other materials to transfer, or if the future carbon-based costs are reduced to an acceptable level, In short, there must be a head start.

“It’s the age of fairy children…then I’ll be relieved.” Tang Wenjie chuckled upon hearing this: “At least my job will be saved.”

Su Yuanshan: “…”

Qin Weimin: “…”