Chapter 104: Take advantage

Zhan Kaihe's emotions were unconsciously driven by Liu Fan, but he calmed down quickly, "This is indeed a wonderful vision, but this matter cannot be resolved overnight."
"Although it does take time, this time may not take too long." Liu Fan took out the tablet. "With the current mainstream deep learning algorithms, if you want to carry it on the current general-purpose chip, there will be huge performance waste and matching problems. The first solution to this problem is to make a breakthrough in architecture.
I didn't go into depth when I talked to them about the chip architecture, but in front of the two, I think I can show what I really want to do. "
Wu Jishui and Zhan Kaihe approached the tablet for a while, and Zhankai Channel, "Dynamic discrete shunt operation."
"Yes, this is the core of the entire algorithm architecture." Liu Fan leaned forward a bit, and Zhan Kaihe noticed the most important places at first sight, which made Liu Fan impressed him very well.
Zhan Kaihe's expression looked calm, but he nodded slightly, and continued to look down at Ji Mushui. It took a long time before he took off his eyes and rubbed his eyes. "No wonder Ji Lao always praises you in front of us. From theory In fact, this is indeed an optimization of the existing FinFET process, which can also greatly reduce the power consumption of the chip. Unfortunately, there is no experimental data.
In addition, there is still no way to solve the problem of matching the depth algorithm with the general-purpose chip, but it is already a great thing for you to have such a technology invention. It is indeed a legend of China. "
"This solution does not solve the problem of how to produce AI general-purpose chips, because it is designed for AI-specific chips. I have actually made a theoretical solution about AI general-purpose chips. Two people can see it." When they asked them to make AI general-purpose chips, they decided to reveal something a little deeper.
Zhan Kaihe and Ji Mushui were a bit surprised. After Liu Fan opened another file in the tablet, the two concentrated again. Ten minutes later, Zhan Kaihe's face was almost stuck on the tablet, apparently without the calmness before, "Xiao Liu, where did you get this inspiration?"
凡 Liu Fan's body moved forward a little bit again, "In fact, the principle of this thing is to learn the rivers on our planet flow into the sea, so I named this architecture called the living water architecture."
Zhan Kaihe nodded at high frequency. "Unfortunately, there are still too many problems to be solved. It's just a matter of materials ... Ji Lao, is there a nano-scale material that can withstand this curvature?"
Wu Jishui shook his head. "In my impression, it is not, not just the material. He should not be able to do such a hooked link."
"In fact, it is not difficult to implement this link method, long-term training should be no problem, as long as a suitable tool can be made." Liu Fan explained a sentence and demonstrated a rough process by hand.
Tong Jimushui and Zhan Kaihe looked at each other. Jimu Waterway, "In general, the core problem of this architecture should be materials. If the material problems can be solved, it is still worth exploring."
"So this is the chip design architecture that I think is most likely to meet the AI ​​general-purpose chip. The development of materials has been very rapid in the past two years. I think we can be optimistic about the future of materials." Liu Fan did not put Material solutions are also available, and it is not necessary now. "And most importantly, this architecture can be split in the design of some special-purpose chips, so that the current materials can fully drive this design.
So my current plan is this. In three steps, the first step is to use the improved architecture to quickly optimize the existing AI-specific chips, and I personally think that the best word is mobile phone chips, because this piece of Hisilicon The experience is very rich, and the user coverage rate of the terminal products that are effective and fast is high, and it can also attract wider social attention.
The second step is to try to design and produce AI special chips using live water architecture and conduct market inspection. Once successful, the efficiency of AI chips will definitely be significantly improved.
The third step is to find materials and work hard to develop the AI ​​universal chip we want. "
After Liu Fan said, Ji Mushui and Zhan Kaihe were in deep thought at the same time. After a while, Ji Mushui asked Zhan Kaihe, "Kaihe, do you think this thing is reliable?"
Zhan Kaihe took his eyes down and said solemnly, "Everything is now at the theoretical level. I think the rational analysis is too risky and the cost of failure is too great."
The car fell into silence, and after a few seconds, Zhan Kaihe said, "But I don't know why. I especially want to take a risk."
Liu Fan and Ji Mushui looked at each other, then appeased, "In fact, because we go in three steps, the whole trial and error cost is not as great as imagined ~ EbookFREE.me ~ If the first step can not go forward There is no second step.
And the success of each step, I think the key is not the upgrade of chip performance, but the improvement of our research and development capabilities and process level, even if we can not break through in the final production of the chip, but the experience and progress gained in the process are believed to be Chip development is also a huge asset. And I feel that at least in the optimization of mobile phone AI chips, I am confident enough. "
As soon as the utterances fell, the car stopped right in front of the hotel. At this time, Ji Mushui patted Zhan Kaihe's shoulder and smiled, "Kaihe, it's time to eat."
Zhan Kaihe's heart moved, and he understood the meaning of Ji Mushui.
Uh ...
Two days later, in the Mulong Technology conference room, Ji Mushui and Zhan Kaihe were sitting on one side, Wang Tao, He Tingbo, and Zhang Rujin were on the other side, and some backbones brought by various companies were sitting around the conference room.
The eyes of these people are now focused on Liu Fan who is standing in front of the stage.
凡 Liu Fan gave a detailed explanation of the entire plan through the PPT. This time, the division of labor, model, budget, and future planning are all clearer than when talking to all parties in private.
Everyone sitting under the stage listening and suddenly realized that a problem, this time the chip project from the front-end architecture design to the back-end process improvement, the core technology is actually all brought to you by Liu Fan?
三 All three parties suddenly felt inexplicable as if to take advantage. At the same time, everyone couldn't help wondering, since Liu Fan had this ability, why would he want to cooperate with everyone? Can't we just hire people to form a team? This is not reasonable.
大家 In the process of everyone's doubts about this issue, Liu Fan finally talked about the most sensitive topic today: how to cooperate ...