As soon as Qian Ruiya's resolute words fell, before Su Xiaomu had time to answer, Su Ho rushed over.
"Sister Qian, won't you think about it again?"
Then he added, "Build another laboratory, if the investment scale is relatively large..."
Qian Ruiya looked at Su Ho with a half-smile, picked up the teacup and drank tea without saying a word.
The atmosphere in the living room suddenly became very awkward.
Su Ho's face turned red and she lowered her head to drink tea.
Everyone is a vixen, so don’t hide it from anyone. The little Jiujiu in your heart has long been... revealed.
As Su Ho's best friend, Wan Dai very rationally helped to continue the operation at this time.
"Sister Qian, when we build a new laboratory, do we need to make a long-term plan?"
It abruptly broke the awkward atmosphere.
Get back to business.
Qian Ruiya blew a few tea leaves floating in the teacup and took a sip of tea. "Xiaomu, tell me, we don't have any written things like investment plans to show off."
In the past, before Su Xiaomu opened his mouth, even in a more embarrassing situation, Su Ho would interrupt and say something like, "Don't be shy, please speak up."
But Soho didn't say that today.
Her identity is different from before. No matter how worried she is, no matter how familiar she is with Qian Ruiya, she still has to help Su Xiaomu save face when she goes out.
Su Xiaomu did not speak immediately. After pondering for a moment, he collected his thoughts before answering.
“I didn’t have much idea about the computing resources of supercomputing before, but now I know very well how many simulation operations and experiments can be done with one trillion times of computing power per second...
Let me first talk about the concept of my plan.
According to the current development and some of the software and hardware that must be completed in the plan, there will be a shortage of at least 10,000 trillion calculations per second in the next one to two years."
Su Xiaomu explained methodically.
"The first is the complete autonomy of Android. This is a large ecosystem. It does not mean that every core module can be replaced and launched first for cooperative testing and then iteration like the file system.
The Android file system has certain particularities, because Android's own software ecosystem is too complex and requires limited permissions and resources to be controlled.
The industrial simulation testing and iteration of the Android overall module alone requires at least 20,000 trillion calculations per second. Of course, this process is difficult to carry out simultaneously."
After a pause, Su Xiaomu continued.
"This is the first; the second is the imminent storage operating system. Because of its professionalism, the size of the operating system is not large, and the overall size may be less than 300m.
The cumulative computing power required must exceed 8,000 trillion times per second, and all modules of the storage operating system are processed simultaneously, which is not as convenient as the Android ecosystem."
Su Xiaomu took a sip of tea and continued.
“Finally, there is the exploration of biological information at the hardware level. This is the future development trend. It is best described by our country’s favorite saying, ‘overtaking on a bend’.
DNA storage technology has been proposed for more than half a century, but until now there has been no huge breakthrough. Giants such as Microsoft are still in the exploratory stage.
Of course, they are trying to do both, and I personally prefer to directly develop DNA storage technology. To put it simply, we don’t have so many worries.”
"The computing power required for this is huge. If it is expected to expand 10,000 teraflops of computing resources, at least 1,000 teraflops of computing power must be reserved for this project in the long term.
"
In the end, Su Xiaomu concluded, "Of course, maybe you heard a little vaguely, this piece of work requires tens of thousands of trillions of computing power, why did I only say 10,000 trillion times.
It’s because I just made a mental calculation, and based on the principle of not wasting resources, a computing power of 10,000 trillion times per second is the most appropriate.”
Qian Ruiya: "..."
Su Ho: "..."
Wanda: "..."
Du Ruo: "...Is Brother Xiaomu so powerful?"
Qian Ruiya thought for a while and said, "Both the storage file system and the Android file system have fixed the first version. I don't think there is any problem with this, but the final DNA storage technology..."
After a pause, Qian Ruiya finally stopped beating around the bush.
"Are you sure we can develop it?"
In fact, Qian Ruiya could have chosen a more tactful approach, but since she is from her own family, she would not do this.
There are many things to consider here, not just the problems of this laboratory, but also the problems of endless investment that can be imagined in the future.
Su Xiaomu smiled, "Let's go to 99%, leaving one point for heaven."
With the knowledge materials provided by the Xiaolaji system, Su Xiaomu would definitely be able to get it out if he was willing to continue exploring in depth.
There is already a lot of knowledge stored, and what is needed is to use experiments and exploration to advance. If it doesn't work, the small garbage system should at least produce the last relevant flame.
After all, fundamentally speaking, Su Xiaomu has some understanding of the small garbage system. The small garbage system has obsessive-compulsive disorder and is not very good at hiding things.
Qian Ruiya then asked, "Then your suggestion is to build a large supercomputing center directly, or to build it dispersedly?"
"I think it's possible to build only one laboratory, but the supercomputing center will be built in phases, one phase at a time. If we directly build a large supercomputing center, the cost may be very high," Su Ho said.
Su Xiaomu nodded, "I agree. Currently, there are not many single supercomputing centers in the world that have more than 10,000 petaflops. There are only about ten or twenty. If it involves breakthroughs in some important technologies, the cost will be relatively high."
In the end, it was Su Xiaomu who had the final say and divided the construction of supercomputing centers into five groups. The first phase will be built with 1024, the second phase will be built with around 2000, and so on. There will be a certain buffer in time and the pressure on funds will be lower.
.
The next thing has nothing to do with Su Xiaomu. Su Ho and Wan Dai will complete the location selection of Mu Shu's second laboratory based on the best combination...
…………
…………
"Xiaomu-san, do you want the difficulty level to be higher today?"
On the afternoon of Wednesday, Su Xiaomu took Shang Siyu and Chen Jiming, who had slowly caught up with the idea, to continue the challenge. Shang Siyu suddenly suggested.
Chen Jiming was stunned for a moment and expressed his support, "Yes, Su Xiaomu, if the difficulty of our training can be slightly higher than the differentiating questions in the finals, the finals will be stable."
Su Xiaomu thought for a moment and nodded.
Su Xiaomu, who gradually learned to distinguish the huge algorithms at work from the algorithms that solve practical problems encountered in training, would have been able to challenge higher difficulties if he had not been afraid of overthrowing others' achievements.
Of course, this does not mean that the finals championship has been locked in advance, because no one is sure how difficult this year will be.
I quickly finished selecting today’s 18 questions and got into the mood.
