Author | ZeR0
Edit | Shadow of indifference
Chipeast reported on September 23 that the 2023 Intel ON Technology Innovation Conference opened in San Jose, California, on September 19, local time. On the first day of the conference, Dr. Wang Rui, Senior Vice President of Intel Corporation and Chairman of Intel China, had in-depth exchanges with media such as Chipdong on Intel's technological progress and strategic layout. Wang Rui said: "Today we can see that our milestones are what we say. The biggest confidence for me is that Intel has recaptured its own progress. ”
"Intel's competitor is itself. The same is true of any large company, when you keep moving forward, you often trip yourself; It's a difficult journey to recognize the problem and then find a strategy to change it and start actually turning the engine back up. She believes that Intel's driving force and supporting role for the "core economy" from the bottom is one of the few in this world, as long as the execution is implemented, there must be opportunities, in China, opportunities are never a problem, the problem is how to do so as not to let opportunities slip away.
In the process of this exchange, Wang Rui explained to us how Intel and the "core economy" are connected, and Intel's specific actions to achieve "making AI ubiquitous". In her opinion, Moore's Law has been pushing forward and Intel's determination has created that artificial intelligence (AI) can really "take off" today. As long as there is demand, it is not a way to open up the road to Moore's Law, but when a thing "hits the wall", and then find a new path to innovate, which allows all semiconductor companies, including Intel, to continue to innovate and achieve exponential growth in computing power.
▲Dr. Wang Rui, Senior Vice President of Intel Corporation and Chairman of Intel China, communicated with Chinese media during the Intel ON Technology Innovation Conference
First, talk about AI: AI creates a good carrier for the economy, and chips become the basic needs of the future economy
The improvement of computing power has allowed many human imaginations about science fiction to be realized. "The driving force of the huge economic system, just as the driving force was crude oil in the early years, this time it started with chips. Chips have become a fundamental requirement for the economy of the future. Wang Rui said.
According to her sharing, AI has created a very good carrier for the entire economy, the chip semiconductor industry size is about 574 billion US dollars, driving the global technology economy of about 8 trillion US dollars, and driving the scale of the digital economy of 38.1 trillion US dollars, of which China has 6.9 trillion US dollars of digital economy scale.
During the conference, the slogan "AI is everywhere" could be seen everywhere, which also conveyed Intel's firm determination to fully invest in AI.
Wang Rui said that the influence of AI is everywhere for large-scale language model training, small-scale and inclusive. From the perspective of Intel's own technology deployment, there are AI applications from the chip to the application layer, whether it is in mobile phones, client PCs (personal computers), edge devices or cloud data centers, no matter what kind of applications are done, AI can be available anywhere. When there is a certain amount of computing power, Intel builds AI capabilities into the chip, so that users can apply these AI capabilities.
From the client side, AI PC is the next generation of PC innovation, and the future way out of PC is on AI PC. Many of the demos shown by Intel at the ON Technology Innovation Conference were made by Chinese teams, and Intel and local partners in China have a lot of cooperation to develop the next generation of PCs. Moving AI to PCs, its applications are just beginning, and there are many applications imaginable.
In the data center, according to the requirements for computing and the scale of the model, users can directly use Intel Xeon Scalable processors with sufficient built-in AI capabilities, and can also use Intel's AI accelerators, and for larger scale requirements, they can extend from one rack to the entire cluster. No matter what kind of requirements, AI is built into it, with different computing power and different architectures to support different needs.
Second, talk about manufacturing: to the Intel 18A node, Intel will return to the leading position in the process
In terms of manufacturing, Intel opens up its advanced process capabilities to the world, not only for its own product manufacturing, but also through the foundry business for enterprises that only do chip design to provide production services.
In response to the problem that the pace of CPU iteration in Intel's data centers has become faster, Wang Rui said frankly: "If we have been steadily moving forward to do it, there would not be the current situation." Now we have to chase, when chasing, you can only run fast, and running fast in our industry is to iteratively accelerate. ”
In fact, she said, not only has Intel itself accelerated, Intel has worked with customers and industry to help them clarify system-level thinking and plan and implement their roadmaps based on Intel's products.
"Once we get to the Intel 18A, I believe we are back at the forefront of the process." Wang Rui said, "At this time, we can have some more steady, step-by-step rhythm, but now in the time of change, we are willing to go through this stage of change with our industry partners." ”
Third, talk about confidence: The client has many Chinese partners, and the data center architecture is flexible to meet the needs
Talking about why he has confidence in Intel, Wang Rui said, first, the grand digital economy trend is carried on the chip; Second, Intel is not a company that does one or two things, and its scope includes both cutting-edge scientific research and real things from transistors to software architectures, from data centers to edges and clients.
On the client side, Intel has a lot of Chinese partners who are working on AI PCs to explore how to apply generative AI applications similar to ChatGPT.
In terms of data centers, Intel will release processors with 288 cores; In terms of computing, P-Core (performance core) and E-Core (energy efficiency core) can work together with the software system to achieve dynamic switching between high-performance requirements and energy-saving requirements. In the future, when the data center needs high performance, it will automatically switch to the performance core; When not needed, it can be transferred to the energy efficiency core.
Data centers attach great importance to environmental protection and energy saving, for which Intel can dynamically switch every watt of power consumption, and the architecture can be simplified, without the need to jump back and forth from different architectures. Intel's open oneAPI programming model builds a bridge between developers and the underlying hardware, whether the underlying is CPU, GPU or other hardware architecture, developers can develop in their familiar programming languages, and oneAPI will automatically "translate" it into a language that the corresponding hardware can understand.
Fourth, talking about automobiles: China's new energy development is amazing, hoping to support the future automotive computing needs
In China, the Intel Greater Bay Area Technology Innovation Center showcased a series of products and solutions that Intel and partners have built an ecosystem with, including liquid-cooled data centers, various innovative PCs, various edge applications, and cars that may become "mobile data centers" and "mobile AI centers" in the future.
Talking about the automotive opportunities that Intel values, Wang Rui observed that the development speed of China's new energy vehicle industry is very amazing, from Intel's point of view, new energy vehicles on the one hand is automatic driving, on the other hand is in-vehicle entertainment, a large number of applications on the demand for computing power has been greatly improved, so Intel and customers continue to deepen discussions, hoping to make a good contribution and support for future automobiles.
When asked how x86 supports the automotive software ecosystem, she said that technical solutions will never be the same, such as Intel is working with partners to strengthen support for the Android ecosystem, previously through Intel and Tencent joint development, on Tencent App Treasure, everything Android can be supported on the PC. For cars, for Intel's partners, Intel will develop together according to their needs.
Conclusion: ChatGPT application is only an intermediate step, and it is recommended that AI entrepreneurship be seriously considered
At the end of the exchange, Wang Rui also shared her views on ChatGPT and generative AI boom: "The current applications of ChatGPT are only a very short intermediate step, and as we move forward, we will see more and more applications. ”
She suggested that if you want to start a business in the direction of AI, you need to think carefully, what your goals are, what are the applications, if you focus on the underlying technology, you need to think about how to do more and more in the technology, and you need to think about what the business model, target market and customers are, and the core sentence - how to help customers make money.
The truth is very simple, the key is to "be practical", no matter what you do, how to find your own competitive advantage in the chosen field, can help customers or help the target market to achieve these things. "After figuring this out, there are so many applications of AI, and I think I can really use my imagination now." Wang Rui said.