Shulou(Shulou.com)11/24 Report--

CTOnews.com Sept. 20, 2023 Intel on Technology Innovation Conference was held today. Dr. Wang Rui, senior vice president of Intel and chairman of Intel China, was interviewed by the media. The following is a live record of CTOnews.com.

Dr Wang Rui:

We have put forward the "core economy" today. Why the "core" is raised to an economic point of view, and what the connection between this thing and Intel is like? tell you about it from my own point of view.

Today, we talk about AI, about a whole new era, where the underlying things are supported by semiconductors, and everything related to computing needs semiconductors. Whether it's AI or Internet, they fly high and fast, but almost all of these companies are built with semiconductors in the underlying technology.

After so many years of evolution of semiconductors, the determination that Moore's Law has been pushing forward and our determination have made today's AI really "take off". As long as there is demand, we are not a way to break through Moore's Law, but when something "hits the wall", we find a new way to innovate.

Quantum physics, Glass Substrate (glass substrate), 3D packaging, which were previously unthinkable. So companies like Intel, not just Intel, but all semiconductor companies, can keep innovating and allowing computing power to grow exponentially.

How much human beings start with mythology, from prophecy, and finally read those fairy tales, and many of the science fiction stories written 100 years ago have come true today. Sometimes I wonder if these very creative science fiction writers have created the technological world, because people invent and create based on them, or vice versa. Sometimes it is really intertwined, because there is imagination, so people have to achieve it. With the improvement of computing power, human imagination can be realized.

In an era like today, AI has created a very good carrier for the entire economy, and these are some of the numbers you hear today. The size of our chip industry is about $574 billion, and it has promoted the scale of the technical economy of $8 trillion and the scale of the digital economy of $38.1 trillion. In China, there is a digital economy of US $6.9 trillion, accounting for 41.5% of the total GDP. The driving force of a huge economic system, just as crude oil was in the early years, this time started with chips. Chips have become the basic needs of the future economy.

So we were just talking about, why have confidence in Intel? First, the trend of the grand digital economy is carried on the chip. Second, Intel is not a company that does one or two things, but from cutting-edge scientific research to actually from transistors to software architecture, from data centers to edges, to clients. Think about how many generations the PC has changed. The first computer was a big room, then it slowly became a desktop computer, and then it became mobile. The early mobile device was like a thick brick, and people thought it was amazing at that time. Today, we have a cooler PC, we have seen it on the stage today, and it is not the only one, we have a lot of partners, and there are many partners in China, how to bring AI directly to the PC application, the magic is that it does not need to connect to the network. What we are doing is AI PC, how to apply generative AI applications like ChatGPT. I believe that PC will help us on another level in the future. In the client business, we do not know how many generations of change, and then change, I believe it will go further and further.

On the data center side, you can see our roadmap, and next time we will release a processor with 288 cores. In computing, P-Core (performance core) and E-Core (energy efficiency core) can work together. In the future, when your data center needs high performance, it will automatically switch to the performance core, and it can switch to the energy efficiency core when it is not needed. For future applications, sometimes you need strong computing power for different workloads, and sometimes you need energy-efficient computing. When we put E-Core and P-Core together, we can build the software system, and computing can dynamically switch between high-performance requirements and energy-saving requirements, which can inspire how many developers and applications?

As we all know, data centers attach great importance to environmental protection and energy conservation, and they don't know how much energy a data center needs. We can dynamically switch per watt of power, and then the architecture can be simplified, because you don't have to jump back and forth between different architectures. Intel's architecture helps you do it. Let's go to the upper layer to talk about our software ecology. Various hardware architectures add a layer of oneAPI to it. This is our open developer toolkit. It is so that developers do not need to know what the underlying hardware is when they develop on it. If you have learned a little bit of programming, the earliest time we have to do assembly language programming is to write machine code, 0 and 1, you need to know how to string 0 and 1, it can directly drive the 0 and 1 of the transistors below us to perform its functions. This kind of programming is very difficult, and if you write a 0 and a 1 wrong, you have to proofread them one by one. Later, it gradually became C language programming, etc., and after it became more and more advanced, it gradually became a language that human beings could understand to do programming. So, in fact, the developer platform we are doing today is very similar to the evolution of the original programming language. it builds a bridge between the hardware under the computing platform and the developers-- as long as you understand the development language. let it perform this function, write it out, and our background infrastructure will help you translate it and transfer it to the entire hardware level. In this way, the whole ecology will be open and your computing power can be fully utilized. This is why such a platform is so important to developers.

Then there is the connection to the network. Intel is also trying to connect all these things. So we've been talking about the five super technological forces, ubiquitous computing, ubiquitous connectivity, infrastructure from the cloud to the edge, AI, sensing and sensing, and getting stronger and stronger.

And then Intel has a manufacturing process. Intel does not only design, but also its own manufacturing process. In the past, we only did it for ourselves, but now we open our manufacturing process to the world, that is to say, we also have to do contract manufacturing, not only for ourselves. It also provides services for enterprises that want to use Intel contract manufacturing business to make their own chips.

In the middle of this, Intel continues to expand. When we finished the PC, we found that there is more and greater growth, how do we use our own ability to achieve these growth, we began to transform step by step.

People often ask me, who is Intel's competitor? I keep telling you that Intel's competitor is itself. The same is true of any big company, when you keep moving forward, you often stumble yourself. It is a difficult process to recognize the problem, then find a strategy to change it, and start to really turn the engine back on.

When Pat comes back, we can see today that our milestone is true to our word. What gives me the greatest confidence is that Intel has put its progress back into its own hands. The external environment is ever-changing, and we have so many advantages that as long as we put our executive power into practice, we will certainly have a chance. In China, opportunity is never a problem, but what we can do to keep it from slipping away. For example, the AI we are talking about today, how many opportunities have you seized? We want to optimize with our customers and ecosystems.

In that case, today we talked about whether we need to build a big language model, need to do training, reasoning, or we are talking about making AI ubiquitous, you don't need to build different architectures. In the whole open architecture, almost all workloads can be implemented on this.

In China, in Intel's innovation center, you can see how we work with our partners to build ecology. We work with partners to build liquid-cooled data centers, or a variety of innovative PC, or edge of these applications, or even cars. It can be said that the car of the future can really become a "mobile data center", a "mobile AI center", which requires more and more computing power. Now there may be thousands of chips in a car, with different functions, each doing a little bit of small things, so it is not so demanding on computing itself. However, in the future, if we can integrate these thousands of chips into a more high-speed and efficient computing system, turning thousands of chips into hundreds, Intel also has a lot of exploration in this area.

Intel for the "core economy" from the bottom of the driving force and support, I believe it is only a handful in this world.

Reporter: today around AI PC, I would like to ask, because Intel's ecology has always been very strong, in the ecological construction of AI PC, what is different from the traditional PC ecology, in the new ecology, what kind of plans does Intel China have? What kind of cooperation will there be in the future to seize the opportunity of this new era?

Dr Wang Rui:

Very good question. In fact, the innovation of PC is generation after generation, AI PC is the innovation of the next generation of PC, and the way out of PC in the future is on AI PC. Many of the Demo we see today are actually made by Chinese teams. We have a lot of cooperation with local partners in China. In the future, you will find that this is really a very good opportunity for innovation. We have a lot of partners in China who are working together to develop the next generation of PC, which you will see in the near future. And I think it's just getting started, and once you move AI to PC, there are a lot of apps to imagine.

Reporter: Hello, Chief Executive, Meteor Lake is also a major innovation in the past 40 years, the structure is also the use of the latest separate modular structure, including NPU, AI chips. In fact, we have also seen a downward trend in the consumer PC market as a whole, while the growth of the data center is relatively strong. Intel first used AI in the consumer market. Do you think, from the point of view of users or corporate partners, which part of Intel's growth may be more important?

Dr Wang Rui:

That's why Intel has different divisions. Which one do you think is more important? It's all important. But its corresponding workload and application scenario are very different. And I believe that in the future, from the perspective of PC, or AI in the client, cloud, and data center, there should be an interoperable relationship, and what can be done should be done on the client side and the edge side as far as possible. When the cloud is needed, it can also be applied on the cloud through a strong connection. So the whole thing has to be connected. The reason why we first released the AI application on PC today is that now, through our release of these products, you can already do this application, and there are a lot of innovations in this area.

There will be more and more data centers in the future. In July this year, Intel officially launched Intel Gaudi2 in the Chinese market, and now we are constantly tuning.

As of today, Gaudi2 has already been released. In the future, Gaudi3 will come out soon, and there will be Falcon Shores. So from Gaudi2, Gaudi3 to Falcon Shores, it has become the roadmap for the development of accelerators and AI computing in the future, plus the most powerful AI applications. So when we say that AI is ubiquitous, it is on the client side, on the edge, and on the cloud. With large language model training, small-scale and inclusive, the influence of AI is everywhere.

Reporter: you just mentioned a point, when the pace of Intel slows down, the lower reaches of the industrial chain is more confused. When your pace suddenly becomes faster, they also have the same confusion. For example, our fourth generation is extremely strong, to the fifth generation, there may be only one year in between. It takes you a year to develop, consumer products are acceptable, but in the enterprise market, it is unacceptable, because I invest a lot of money in a data center, maybe not a year, your CPU has been updated, and the performance and power consumption have been greatly enhanced, so what should I do at this time? I believe this is an issue that many domestic enterprise users or enterprise server manufacturers are more concerned about. What do you think of this problem?

Dr Wang Rui:

If we had been moving forward steadily, this would not have happened. Now we want to chase, when chasing, you can only run fast, and fast running in our industry is to accelerate iteratively.

In fact, it is not only Intel itself that has accelerated its pace, but we are working with our industrial partners to help them clarify their thinking at the system level. We have worked with our customers and industry to help them plan and implement their roadmap based on Intel products. As Pat said, once we get to Intel 18A, I believe we have regained our leading position in the manufacturing process. At this time, we can have some more steady, step-by-step pace, but at the moment of change, we are willing to go through this phase of change with our industrial partners.

Reporter: I would like to ask a question about cars, because I remember that there is an intelligent cockpit in Shenzhen Innovation Center. Do we have anything new to introduce in this area? I'd like to know, what opportunities do we value most in this area?

Dr Wang Rui:

I think in terms of new energy vehicles, the development speed of China's industry is very amazing, and all kinds of manufacturers are constantly increasing their efforts.

From Intel's point of view, new energy vehicles are self-driving on the one hand and in-car entertainment on the other. When you need so many applications, what is the need for its computing and its capabilities? Compared with the original very small neural computing unit (NPU), the demand for computing power has been greatly increased. So our customers began to come to Intel, saying that now I have too many things to ask for computing power, Intel can help us. So we began to discuss with our customers, and these discussions are still deepening, and we hope to make a good contribution and support for the future cars in this regard. I hope you can hear our specific release in the near future.

Reporter: in the automotive software ecology, Android has certain advantages, if it is x86, how to support the automotive software ecology?

Dr Wang Rui:

The technical solution is always more than the same. I wonder if you have seen it? Recently, on Tencent YingYongBao, everything on Android can be supported on PC, which we developed together with Tencent. From an ecological point of view, we are constantly working with our partners to support Android. On the other hand, Windows system, of course, has its own set, for cars, for our partners, we will develop together according to their needs.

Interviewer: I have a problem. I saw today that AI is ubiquitous. We are in hardware and software, from cloud to end, from training to implementation, including optimization of large models. I think the bottom two layers are Intel's most solid ecology, but in the upper two layers, I think there may be a lot of opportunities, which is precisely the very active area of innovation and entrepreneurship in China. First of all, I would like to ask you to talk about it, because many startups are training large models of reasoning, can you talk about your views on such a trend from your point of view? In addition, how can we continue to strengthen our ecology on these two layers?

Dr Wang Rui:

AI is everywhere. From the point of view of Intel's own technology deployment, there are AI applications from chip to application layer, no matter on mobile phones, PC, edge devices or data centers. No matter what kind of applications you do, no matter what kind of applications you do, whether medical or educational, there can be AI anywhere. When we have a certain amount of computing power, we build AI capabilities into our chips, and you can apply these AI capabilities. For example, Stable Diffusion, which is a very good application on the client side. In the data center, according to your computing requirements, the scale of this model, you can directly use our Xeon, it already has enough built-in AI capabilities to support. If you need more, you can use an accelerator, which can extend from one rack to the entire cluster. No matter what kind of needs you are, AI is built into it, according to your needs, with different calculations to support, different architectures to support. So let's go back to the hybrid architecture, XPU.

For decades now, everyone has been promoting the development of AI, and now it is time for many dreams to come true. The other is the craze of ChatGPT. In my opinion, the current applications of ChatGPT are only a very short intermediate step. As we move forward, we will see more and more applications.

If you want to start a business, you need to think carefully, what is your goal and what is the application? If you are focusing on the underlying technology, think about how you can get deeper and deeper in the technology. You also need to figure out what the business model, target market and customers are, and go back to that sentence, how can I help my clients make money. After thinking this through, AI has too many applications, and I think we can really use our imagination now. But be practical, no matter what you do, and how to find your own competitive advantage in your chosen field, whether horizontally or vertically, that can help your customers or help your target market achieve these things. In fact, there are many reasons are still very simple, return to nature.

Host: thank you very much for your time and thank you all. This is the end of our communication today.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.