AI, a highly discussed topic this year, whenever we mention the wonderful use of AI, most people ignore the success of AI, in fact, it is closely related to the rapid development of our chip industry in recent years, interestingly, chips have contributed to a new generation of AI, and AI has contributed to a new generation of processors.
December 15th,Intel officially released the fifth-generation Xeon processor and Intel Core Ultra series processor in China, these two products that have caught up with the tail train in 2023, what is the biggest difference between them and the previous processors?
The answer is: AI.
Let's take a look at the fifth-generation Xeon processor first, for ordinary users, this series may be a little unfamiliar, but in the server field, Xeon was once the world's most powerful processor series, carrying half of the human Internet, and now, Intel has given it more missions on the fifth-generation Xeon processor.
Looking at the performance of the processor alone, it is generally a regular upgrade, but in terms of features, the fifth-generation Xeon processor has made many upgrades, such as:Each core is built with an Intel AMX Accelerated AI ModuleNew scalable security features and protections based on advanced cryptographic computing technology can better protect the AI models running on the server from unauthorized intrusion and tampering.
According to the data given officially,With a 21% improvement in base average performance and a 42% improvement in AI inference performance compared to the previous generation, Xeon 5th Gen processors can handle demanding AI workloads without the need for a separate AI accelerator thanks to the built-in Intel Advanced Matrix Extension.
However, the fifth-generation Xeon processor has little to do with ordinary consumers, and is more used in enterprises and data service centers, and domestic leading cloud service companies such as JD Cloud and Alibaba Cloud have cooperated with Intel to deploy and apply the fifth-generation Xeon processor to strengthen the AI processing capacity of the data center and improve the overall efficiency.
Next, let's take a look at today's protagonist, the Core Ultra series processor, which is Intel's first consumer-grade processor to support the deployment of AI models on the device sideIt has a built-in professional NPU module developed by Intel specifically for AI processing needs, which can significantly improve the AI performance of the processor, and allow the Core Ultra to carry the operation needs of AI models with high data volume alone.
When the Core Ultra series was announced, Wang Zhicong, vice president of Intel's marketing group and general manager of China, said this:
"AI has gradually entered people's lives, and the era of intelligent, diversified and open AI has arrived. In the future, AI will be deployed in a hybrid manner and flexibly deployed at the cloud edge. The cloud platform will continue to provide the strongest cornerstone for AI training, and AI PCs with stronger computing power will enable everyone to experience the convenience brought by AI. ”It is true that the time for AI models to appear in front of us is still very short, but I believe we have all seen the development speed of AI models in the past yearFrom simple dialogues to complex data collection, programming, and planning, from simple imitation paintings to whimsical works of art, from simple image generation to complex 3D model animations, the capabilities of AI models are advancing at an astonishing rate. The birth of the AI model may be a key for human beings to free themselves from complicated work, and the popularity of AI models is directly related to whether AI can bring us more universal value, and the task of the Core Ultra series is to let everyone enjoy the dividends of the AI era.
Therefore, we can see that many of the functions of Core Ultra are actually optimized and upgraded around AI, and the NPU performance of the Core Ultra series can support up to 20 billion AI models for local deploymentThis level of AI model has been able to meet the daily use in most cases, and functions such as dialogue, data sorting, Q&A, and Wensheng diagram can be completely generated by local NPUs without using the cloud AI model.
Although the device-side AI model does not have the super-magnitude of the cloud AI model, it has higher security, so that it can be adjusted according to the user's own needs in practical applications, and there is no need to worry about the leakage of sensitive information such as personal privacy information and confidential documents during use, further strengthening the security of the AI model.
In addition, the device-side AI model can be used in a network-free state, making the use scenarios it can cover a wider range of scenarios, and at the same time, the AI model can be called at any time and can be customizedPerhaps the personal AI butler in science fiction movies will become a reality, and with the blessing of the AI model, the PC will also become more "human".
The concept of AI PC was proposed by Intel CEO Pat Gelsinger in September this year, and he believes that in the next few years, 80% of the PC market share will belong to AI PC. With the help of AI, PCs in the new era will be very different in terms of performance scheduling, ease of use, and experience.
We've been talking about the AI performance of the Core Ultra series for a long time, but isn't there any other improvement?
Of course not, the Core Ultra series has also achieved significant improvements in traditional performance and energy consumption ratio. As the first processor built on Intel 4 process technology, the Core Ultra series has a 36% increase in battery life compared to the previous generation, and also has stronger performance.
The Core Ultra series is up to 55% more efficient than any other professional software, and the Core Ultra has been significantly upgraded in terms of graphics performance. In the field test, the performance of the core graphics card can not only support 120 frames of "CS:GO2", but even play "Cyberpunk 2077" at 60 frames, and in the comprehensive test, the core ultra performance of the core graphics card is increased by up to 100%.
As you can see, the upgrade of the Core Ultra series is perfectCPU, GPU, and NPU have been fully upgraded, in exchange for a new series of Intel, in the future, the performance of the three modules will become a key criterion for measuring the processor type, and whether the built-in NPU will also become a watershed between the old and new processors.
After Intel released the Core Ultra series processors, many domestic PC companies such as Lenovo, HP, and Asus have released their AI PC series products, all of which are equipped with Intel's Core Ultra series processors, and some products have opened the pre-sale mode.
Among them,The two products released by Lenovo belong to the two product lines of Lenovo Xiaoxin and ThinkPad, which are aimed at the ordinary white-collar and high-end markets respectively, while ASUS released the Zenbook 14 series ultra-thin and light laptops, focusing on the mid-range market. Judging from the current announcement, there is not much increase compared with traditional PCs with the same configuration, and it is very competitive considering the improvement of performance, battery life and AI model experience.
Moreover, Intel also said that there will be 230+ mobile PCs equipped with Core Ultra series processors in the future, which will gradually cover different price segments and different usage scenarios. With the help of AI, the PC is ushering in a new future as a productivity tool.