"I see," the people at the scene said, their eyes lighting up instantly. They hadn't expected the second-generation AI assistant to be so powerful, with so many practical functions.
However, the professionals at the scene or on the internet were thinking even more than ordinary people.
Because they discovered that this NPU neural network processor could be applied in so many places.
Whether in autonomous driving technology, industrial production, or security surveillance cameras, this NPU processor could be used!
First of all, autonomous driving technology!
What causes the cost of autonomous driving technology to be so expensive?
It consists of sensor costs, host hardware costs, artificial intelligence research and development costs, autonomous driving technology research and development costs, and so on.
Among these, the host hardware cost definitely occupies a very high position.
Why?
Because artificial intelligence interprets the camera's image by analyzing each pixel one by one, and cannot, like the SMM pulse neural network, only pick out the important points and ignore the unimportant ones.
Ultimately, as the number of cameras increases, as the number of radar sensors or other professional sensors increases, the complexity of information will increase exponentially.
This leads to an exponential increase in the computing power that artificial intelligence needs to consume, bringing more expensive host hardware costs.
But now with the NPU neural network processor, even if the number of cameras is increased by more than ten or even dozens of times, it can be tolerated.
Because for the NPU neural network processor, if the information is not important or stimulating, that is, if there is no change in the picture.
Then the NPU neural network processor will automatically ignore the useless data and only record the useful data.
In the end, not only can it reduce the computing power required for autonomous driving technology, but it will also become safer after adding more sensors.
This is the use of the NPU neural network processor in autonomous driving technology!
Of course, in addition to autonomous driving technology, the next is industrial automated production.
If you want to automate production in industrial production, you must always understand the current situation of the items.
Because the NPU neural network processor supports the SNIN pulse neural network, it directly and automatically filters out a large amount of useless data and automatically extracts useful dynamic data.
In the end, the factory can use cameras or sensors with higher image quality, and the number can be installed more.
Anyway, a lot of useless data can be automatically shielded.
The computing power burden brought by the improvement of camera image quality and the increase in quantity is not too great, and the computing power burden will only be lighter than before.
Therefore, cameras and other sensing sensors in the factory can adopt higher standards. This is the use of the NPU neural network processor in automated factories.
The same is true for surveillance cameras after this.
People who have surveillance cameras installed at home must know that the pixels and image quality of surveillance cameras are really poor.
It is completely incomparable to the clarity of mobile phones.
It can be said that installing a surveillance camera at home mainly serves as a psychological comfort. Unless the distance is very close, it is impossible to see clearly.
The reason for all this is actually the problem of computing power and data storage.
It's not that those camera manufacturers don't want to produce cameras with higher pixels and clearer image quality.
The reason is that higher pixels and clearer image quality mean higher requirements for computing power, and this cost will increase.
Finally, the higher the image quality, the larger the volume of stored data will be, and then the memory space will be full after only a few hours. How can security surveillance cameras play a role?
Therefore, the increased computing cost and storage space cost brought about by the improvement of image quality are a big problem.
The result is that if you want perfection, the price of the surveillance camera will be as high as thousands or even thousands of yuan, directly scaring off 99% of consumers.
But now, using the NPU neural network processor can directly avoid this problem and also have many special functions of artificial intelligence monitoring.
Because in the picture of the surveillance camera, most of the picture is actually still, which also leads to a lot of picture data being useless data.
At this time, using the NPU neural network processor can filter out a lot of repetitive and meaningless pictures and only record the pictures that are moving.
In the end, the amount of data that needs to be stored is reduced linearly, and the computing power burden is also reduced, so naturally the image quality can be improved!
It can be said that after careful calculation, there are too many places where the NPU neural network processor can be applied.
It's just a pity that to do these things, you need to be proficient in artificial intelligence technology and also be proficient in the development of the NPU neural network processor.
Ordinary manufacturers without any scientific research strength cannot afford to play with these things.
"Just now, we introduced the special features of the second-generation AI assistant in mobile phone photography.
Next, we will introduce another special feature, which we call 'Frame Generation Technology!'"
"Frame Generation Technology?"
Hearing this technology, the people at the scene showed a confused expression.
In response, Gao Nian gently waved his hand, and then a conceptual diagram of game frame rates appeared behind him.
"If you want to introduce frame generation technology, you must first introduce what game frame rate is.
Game frame rate, as the name suggests, is the number of frames that appear on the screen per second. The more frames, the smoother the picture looks.
If the game frame rate is 30 frames per second, then the game picture will look choppy to our naked eyes. If the game frame rate reaches 60 frames, then the game picture will look very smooth.
If the game frame rate reaches 100 frames, then the picture will look extremely smooth, and the naked eye will not be able to see any abnormalities or inconsistencies.
So 100 frames is a critical point. There is not much meaning in exceeding 100 frames, because if you continue to improve, the naked eye will not be able to detect the benefits brought by the frame rate improvement."
After a pause, Gao Nian continued:
"Currently, the actual frame rate of smart phones around the world is difficult to reach 100 frames or more due to the performance limit of the mobile phone processor.
It can even be said that as the image quality improves and the game size varies, it is actually very good for the game to maintain between 60 frames and 70 frames.
As far as our Geek S2 smart phone is concerned, many new large stand-alone games currently have only about forty or fifty frames, reaching a barely playable level.
Even our Geek S2 smart phone is like this, not to mention other smart phones with worse performance.
Among them, in order to solve the problem of poor game experience caused by too low game frame rate, our Geek Technology Company has also tried every means.
In the end, we developed frame generation technology. This frame generation technology uses the relationship between the previous frame and the next frame.
The AI assistant directly generates a frame between the two, and in the end, our game frame rate will double!
40 frames directly becomes 80 frames, 50 frames becomes 100 frames, and 60 frames becomes 120 frames!
In order to bring game players an extremely smooth experience and avoid the terrible situation of the picture stuttering when playing games!"
"Hiss!"
Hearing Gao Nian's words, the people at the scene suddenly gasped.
They didn't expect Geek Technology Company to develop such a super black technology, which is really amazing.
Those game players' eyes lit up even more, and they called out that they really loved Geek Technology Company. This Geek Technology Company is simply anxious about what they are anxious about.
