Ye Ming knew that Old Tang might have misunderstood - thinking that he wanted to build a supercomputer like Taihu Light.
"My idea is that if conditions permit, we will build a parallel supercomputer with as much computing power as possible." Seeing that Old Tang was still hesitant, Ye Ming gritted his teeth and said, "A top-end GPU parallel computing server will also do."
Professor Tang regained his composure, and when he heard this guy ask for it and lowered it, he burst into laughter: "Please be clear, I want you to make a good plan - at least now you have to make it clear whether it is worth so much."
Money. I mean, compared to us buying cloud computing space."
"Well... it's not safe to buy space." Ye Ming immediately shook his head: "I have thought about it before, and I think it is safer to store the code and data on my own hard drive."
"Then what?"
"Then, I want to accumulate some experience for the task distribution of traditional servers and quantum servers in the future."
As he spoke, Ye Ming told Old Tang some of his thoughts.
To be precise, it was the method he and Yin Ta discussed.
Of course, the people who know the hardware of quantum computers best are Professor Pan and his group. But if the person who knows the software of quantum computers best... it may be none other than Yin Ta.
Under Ye Ming's guidance, she deconstructed and thought about "herself" a lot, and finally came to the conclusion that her ontology should be a quantum computer.
Half a year ago, Ye Ming proposed to Professor Pan a plan to build multiple parallel quantum computing units. They have now completed the production of 50,000 confinement wells - that is to say, the Jiuzhang-3 has already begun.
Have the ability to calculate both types of problems.
But this is definitely not the ultimate architecture of quantum computers, nor is it an architecture that can be promoted for commercial and civilian use.
To achieve commercial and civilian use, it must reasonably allocate and transform computing tasks, and those high-concurrency tasks cannot be handled by the parallel circuits of the quantum computer itself.
As Ye Ming spoke, he drew a four-section framework for Professor Tang on the manuscript paper.
First is the quantum computing unit, then is the parallel array of logic gates, then is the traditional server, and finally is the user terminal.
"For example, there are N simultaneous requests sent to traditional servers. The servers allocate computing tasks, call a different number of quantum computing units according to the complexity of the tasks, and finally return the results to the terminal. In fact, this current cloud computing architecture
It’s similar, except that the final calculation is not the CPU or GPU, but the quantum computing unit.”
"Well, I won't talk about the network response speed - do you have a way to completely divide the classical problem into a quantum problem now?" Professor Tang is not someone who is easily fooled, and immediately pointed out the key to Ye Ming's problem:
"The path Lao Pan and his friends are taking now is what you proposed before. I think that path is very good. As long as we continue to make progress in engineering, not to mention the nanoscale, even at the micron scale, it can also cause quantitative changes to lead to qualitative changes. In fact,
Now Lao Pan and the others have begun to try new parallel algorithms and have made good progress in turbulence problems..."
Ye Ming nodded. He has been communicating with Academician Pan's team this week. Of course he knows the progress...
But he still underestimated the scientific community’s “thirst” for computing power.
What is turbulence?
It is a random movement with a vortex structure displayed by fluid when moving at high speed.
Heisenberg - the main founder of quantum mechanics and the proposer of the uncertainty principle and matrix mechanics - once said that he must ask God two questions, one about general relativity and the other about turbulence.
It is also known as the last unsolved problem of classical physics.
You will know how difficult this thing is.
Fortunately, with the birth of computers, especially supercomputers, scientists have begun to use super computing power to perform fluid simulations - there are no clever ways, but there are always stupid ways.
Therefore, after learning that Jiu Zhang 3 had completed the incredible architectural design, a group of domestic turbulence and meteorology people swarmed in. Lao Pan was also a warm-hearted person, and he really came up with an idea, although he was far from completely solving the problem.
It's still far away, but at least there is hope.
"The idea can only be said to be immature now. Everything will have to wait until the time comes." Ye Ming smiled and said: "Teacher, I don't think we need to wait for others for fifty years, right?"
Upon hearing this, Professor Tang was slightly startled.
After a moment, he laughed: "Yeah! That's right! Okay, but let me remind you that the company will soon pay for the material costs of the foundry, and the workshop will also spend money. I can't give you the money.
Over two million - if you can wait until the end of the year, five million is no problem."
"Besides, you have to talk to Teacher Chen about this matter. It's best to agree to help him take care of some students."
Ye Ming was startled: "I will take it?"
"What else?" Professor Tang saw Ye Ming's confused face, shook his head and smiled: "Now that you have completed your credits, you are just missing a heavyweight paper and results - if you can show it, the school will give it to you in minutes.
professor."
"Uh..." Ye Ming coughed: "Actually, I think this is pretty good."
Professor Tang was silent for a few seconds and then nodded: "I also think this is good. You can break away from the shackles of professional titles, concentrate on your scientific research, and help the company make some money. It is much better than wearing any hat. As long as you show results
Are you worried that you will lose academic reputation and status?"
…
After bidding farewell to Old Tang, Ye Ming went to see Professor Chen again. Sure enough, Old Chen didn't stop him. He just asked him to bring his two AI graduate students with him when he was working on AI... At the same time, Old Chen also gave him a friendly reminder.
, he still has one game to play in May.
When he heard that there was still a game to be played, Ye Ming also played tricks on Lao Chen and asked Lao Chen to help him with the supercomputer, but he was too lazy to bother.
"One Sword Dominates"
In the end, he was scolded out of the office by Professor Chen Xiaofang.
On the way back to the research center, Ye Ming's whole step was in the air.
Playing a competition is such a simple thing... It takes three days to go back and forth, just like going to the capital to experience it and collect a prize on the way.
Back in the office, before his butt was warm from sitting, the ajar door was gently pushed open, and then a girl poked her head out.
"Sister Gu."
The visitor is Gu Qin, a graduate student who newly joined the Lao Tang Group last year.
Since Lao Tang and Lao Chen started their own business, basically all the "old people" in the group have been dragged to "part-time jobs" - these people basically have no worries after graduation, and they don't have any psychological burden when they are "squeezed".
But for newcomers... first of all, there are courses to learn, and secondly, there are no results. Although the salary of Jizhi Technology is not low, it is really not worth the loss to waste one's studies.
Therefore, the main force in scientific research, or the main force in the group, is still this group of newcomers.
For example, this year Lao Tang will recruit four more people.
Since he was the main force in the work, Gu Qin became familiar with Ye Ming.
"Ye Ming, Mo Gu asked you to go there." Gu Qin opened the door a little, revealing the white coat he was wearing.
"Huh? Are you in trouble?" Ye Ming stood up after hearing this.