"I think you've read too many novels! I'm not talking about a blend of flesh and soul! ”

While educating Xiao Huan, Xia Ming also looked away, the legs exposed outside the quilt were slender and firm, and the skin was pink and crystalline, and the lethality was really big enough.

Xiao Huan paused, and then put her head into the quilt, her face was already burning and smoking.

It took Xia Ming a lot of work to calm her mind, and then introduce QNI to her.

"Quantum neural interfaces? What is it for? ”

"Digitizing thoughts and memories? What's the use? ”

"Are you going to turn me into a mecha-hime?"

"I'm going to get into the network like Suzi?"

Xiao Huan, who has no concept in this regard, is a completely curious baby, and the transmission process is intermittent, and it took most of the night to complete.

"It seems like there's something more in my head, like ...... A window, the outside of the window was messy and blurry. ”

"Memory, no, I can't see it."

"Auditory and olfactory information? Not at all. ”

"Think about yourself, well, I understand what you mean, Xiao Huan is very special and unique! Nothing has changed......"

"Ah, I can write and draw things in the window, it's very clear."

After the transmission was completed, Xia Ming let Xiao Huan experience the existence of QNI, thinking that it might be the same as R3AI settling in himself, and there would be great changes to consciousness and senses, but he didn't expect this to be the case.

Yes, R3AI only exists in its own brain, and it only copies and transmits QNI.

Xiao Huan sent a thought: "I probably understand, our current telepathy is a kind of magic, and magic can't be analyzed with our current technology, so you have installed such an operating system in our brain, which can use part of the brain as a computer?" ”

She came up with an idea: "Then why don't you install a QQ anymore, how convenient is it?" ”

I'm really going to think about it.,Install a QQ again.,There's already telepathy and it's too much.,Wait......

Xia Ming patted his head, yes, why don't you install a communication tool similar to QQ on QNI?

The idea sensing with Xiao Huan is a blurred thing, and she has to block the thoughts that she doesn't want her to know. If the communication between the two parties is fixed on the communication tool of QNI, it is both convenient and controllable.

I didn't expect this girl to be quite smart, but in this way, before she is connected to the Internet, she has to build a psychic network first? No, let's take another name.

Xia Ming thought about it and decided to follow the academic norms of scientists, so it was called...... Quantum neural networks, also known as QNNs. As for the communication tool, it must no longer be called QQ, but ......

Unexpectedly, Xiao Huan was also thinking about this problem, so she heard her say: "Let's talk to the spirit, open your heart, and talk together." ”

Spiritual chat? The style of painting has become a fairy tale or fantasy? This can't be done, Xia Ming decided to think again.

After more than a week, Xia Ming took the time to come to the research institute, and he was going to ask Lin Cheng about the chat tool.

It is not impossible to directly decompile QQ, but to transplant it to QNI, it involves the quantum quadruple language developed by Professor Xu and Lin Cheng, and he still does not know enough about some kernels.

"What are you up to?"

In the computer room on the first floor, he saw Lin Cheng, Lawrence and Wu Sihao surrounding a machine, and their faces were all gloomy.

The three of them were mixed with explanations in Chinese and English, and Xia Ming knew that they had encountered a thorny problem.

The Sativa engine, adapted from quadruple graphics technology, is complete, but this engine has two virtual links that make rendering very inefficient.

A virtual link is to virtualize the existing X86 architecture processor into a quadernary hardware architecture, and the quadernary hardware architecture written by Lin Cheng is to let four hardware instructions run in parallel, which requires four cores to run in parallel. If this virtualization link runs on a four-core or eight-core X86 processor, it is not much of a problem, but a single-core, dual-core or triple-core processor needs to be time-sharing, which is too poor in efficiency and will have compatibility problems.

The other virtual part is image presentation, the "particle real reproduction" conceived by Mekrit technology, which is implemented on the Alfalfa engine by decomposing the object into a number of particles according to a standard size particle degree, and then giving each particle some properties. For example, the particles on the surface should have different colors according to the properties of the object, so that when combined, they are more realistic than the objects constructed from the 3D model of the line and surface polygons, and then they can be converted accordingly according to the existing planar dot matrix pixel display equipment.

In Xia Ming's quadruple graphics system, the implementation of the alfalfa engine is very simple, but it is only a conceptual implementation, which is the real vision corresponding to the quantum computing shell and even the Qianyan simulation.

Different from the vision that most people understand, human vision has a sense of space, and every point is accompanied by spatial information. This is not only because people's spatial perception is scattered among the senses and integrated by the brain, which is an organic whole, but also that the spatial perception attached to the visual is also affected by the whole. And because the light received by each optic neuron is not independent, but is mutually validated and compared, the human eye cannot simply be equated with a digital camera.

Although today's display technology, especially VR technology, uses flat dot matrix pixels to simulate a sense of space, so that most people feel that it is no different from the real thing, but if you carefully distinguish it, this sense of space is still fake. Perhaps when technology goes further, just as movies have now generally developed to 48 frames per second, it is difficult for the naked eye to see through the stuttering that this is actually the effect of continuous playback of still pictures, but after all, it is only "showing reality".

Existing display devices can only provide "display reality", and Mekrit must virtualize this real vision and bring it to life by the existing planar dot matrix pixel display device. The simulation of this link, although supported by quadruple technology, is still a seemingly redundant step compared with the technology of directly using plane lattice pixels to present 3D effects, and the efficiency is naturally much worse.

"The processor is the latest i9-8390, 128G memory, and the graphics card is two 980TI SLIs, surrounded by a standard monochrome cube, which is less than 10 frames per second."

Lin Cheng was very dissatisfied with the results, and the particle size they used for modeling was the standard size of the pixel size of a 27-inch 1080P monitor, with a diameter of 30 microns, or 0.03 millimeters, which is almost one-third of the diameter of a hair.

The minimum size that can be seen by the human eye is 0.06 mm at a distance of 30 cm, so this particle size can be "realistic" to the naked eye even on a 4K monitor.

However, this is because the "conceptual reality" of the object is first constructed according to the particle degree, and then it is transformed to present the "display reality" on the display. Even a 3-centimeter-long strand of hair contains 900,000 standard particles, and even if you don't have to fully define each particle through cluster optimization and other means, the amount of computation required is very staggering.

Lin Cheng has always been not very confident in this matter, and now it is almost to the limit: "I still think that this thing is actually worthless, what people want to see is to show the truth, we first create a concept of reality, and then put it on the monitor to become a display of reality, this is a lot of work......"

Xia Ming asked: "The problem is that the amount of calculation is too large, right?" ”

The three nodded, and Lawrence further said: "It really doesn't have much to do with the graphics card, if you use a four-way E7, it should be able to fill the efficiency gap with traditional display technology." But to the market, it's worthless. ”

Of course, the same 3D game, with traditional display technology, an i9 can be played, with the alfalfa engine, you have to play the four-way E7, who cares about you.

Wu Sihao touched on another question: "Why do you always think of games first? Judging from the achievable effect, Alfalfa is a few streets away from traditional technology in terms of fineness and realism, and if you use Alfalfa to do CG, you don't need to render it, which is a revolution for the film and television and CG industry! ”

Lawrence waved his hand and said, "Unless we use Alfalfa to make a CG first, let the industry know that Alfalfa has such ability." The problem is that CG, especially for real-life scenes, has too many problems to solve. Light and shadow effects, physical effects, it takes thousands of people years to add detailed functions to alfalfa. ”

Lawrence understands this business, and the original use of the Alfalfa engine was games, after all, the requirements of the game will be much lower, and the scenarios involved are also very broad, and the details can be fleshed out while using it.

Now that the scope of application is revealed, the question comes back to the amount of computation.

Looking at the ARM server in the computer room, which is put together with a large number of PC and mobile phone chips, Xia Ming had a flash of inspiration: "The amount of single-point computing in Alfalfa is not large, the focus is still too many particles, and the calculation requires a large number of threads." What if we add an additional compute expansion card? Transferring the two parts of the computational cost of particle property definition and particle positioning to the expansion card should not be much worse than the traditional engine in terms of efficiency, right? ”

All three fell into deep thought at the same time.

After a long time, Lin Cheng nodded and said: "Theoretically, I think it's okay, if you use four ARM chips on the expansion card, you can share 60% of the computing effort, and if you use the hardware built-in quadruple core on the expansion card, you can also reduce the amount of computing required for the first virtual link, so that you can share at least 80% of the computing effort." ”

Lawrence clapped: "We can join the cluster virtual in the alfalfa!" The main processor calculates the cluster relationship of the particles, and the graphics card performs the cluster display, which can also improve efficiency! ”

Wu Sihao thought of the most realistic question: "Add an expansion card?" Isn't that the same as the Hanka and decompression cards of decades ago? Who still accepts this in this day and age? And it will cost at least a few hundred dollars? ”

"What everyone really cares about is the effect, not whether to add an expansion card or not", but Xia Ming doesn't think this is a problem.

Speaking of effects, Lawrence and Wu Sihao were relieved, yes, effects!

Mekrit's reincarnation, the Alfalfa Engine, still seems to be subject to the hardware principles of existing display devices, and the displayed picture cannot go beyond the hardware level. However, unlike the linear polygon 3D technology, the core of the Alfalfa engine is real, reproducing the real, and this reality is by no means as simple as the picture seen visually.

There is a direction to solve this problem, but if we start to realize it, we will face another problem, which is an eternal problem: money.

If you want to make an expansion card, you have to find a hardware manufacturer to make a plate, and the follow-up test must also be produced in small batches, which is not something that hundreds of thousands can do.

The first half of the $7 million was given to Tang Xiaotang, and now the money on the books has to be used for follow-up development, and it can't be moved. Xia Ming and Lin Cheng also have private money, but two or three million can't solve the big problem, and it will take a year and a half for the alfalfa engine to see benefits.

Speaking of money, Lawrence subconsciously thought of venture capital, and he proposed to go back to Silicon Valley to find investors. In his opinion, as long as the alfalfa engine has a prototype, it can pull a lot of dollars in Silicon Valley.

"And then we're being kicked out of the board?"

Lin Cheng disagreed, he was most sensitive to this. Xia Ming also disagreed, not wanting to be controlled by capital is one thing, this will also open another door to Yang Yuanchang's unknown forces, and this also involves the technology of quaternary and even quantum computing, and people cannot be let in casually.

Lawrence spreads his hands: "The problem of money has to be solved......"

Wu Sihao opened his mind: "This expansion card, if it can be equipped with a general CPU, so that the old CPU can also come in handy, it should be very attractive to DIY players." ”

Xia Ming slapped his face, good idea!

This project has nothing to do with the alfalfa engine, Xia Ming decided to put it on the side of the cicada, and use the signboard of the cicada to make it a short and fast small project.

Just do it when you think of it, Xia Ming is ready to talk to Li Diluo, it just so happens that he hasn't been to the cicada for so many days, and he has to at least take a look at the new office.

When I went out, I remembered the business of coming to the institute today, and as soon as I told Lin Cheng, Lin Cheng took out a key.

Lin Cheng said: "The quaternary language, the construction of the quantum computing shell, the base compiler, all the information is in Professor Xu's safe. ”

Xia Ming was stunned, this is tantamount to explaining all the family background of the institute to himself?

Lin Cheng glanced at him and said in a low voice, "Now that you are missing, can this thing still be done?" ”

Xia Ming sighed secretly and took the key, at that moment, he felt that this key was not an ordinary Shen, this was not the key, it was the expectation of Tang Xiaotang and Lin Cheng, and it should also be Professor Xu's hope.