Author: Wang Xin **IT Times.
At 3 p.m. on December 15, 2023, laptops equipped with the latest Intel Core Ultra chips will be launched on e-commerce platforms such as JD.com and **. With Intel Core Ultra processors, AI will be the next watershed in the PC industry.
At the same time, the 2023 Intel New Product Launch Conference and AI Technology Innovation Party was held simultaneously, and chips and new products were released simultaneously in China. At the same time as the Intel Core Ultra processor, there is also the 5th generation Intel Xeon Scalable processor, and the common feature of both is that they are both "very AI".
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. Wang Zhicong, vice president of Intel's marketing group and general manager of China, said that this is the first time in more than 20 years that it has released a new heavyweight product for cloud and terminal at the same time.
When the terminal and the cloud meet, the core is more "cool", and the Xeon is "stronger".
AI PCs will become "absolutely mainstream" in four years
December 15th is just the beginning, and the new Core Ultra mobile processor-powered PCs are now available worldwide, with partners such as Acer, Asus, Dell, DynaBook, Gigabyte, HP, Lenovo, LG, Microsoft, MSI, and Samsung expected to bring more than 230 new AI PCs to the market.
Michelle Johnston Holthaus, executive vice president and general manager of Intel's Client Computing Group, said, "AI PCs will gradually become mainstream over the next four years, accounting for 80% of the PC market share. By working with our hardware and software ecosystem partners, Intel is transforming the next generation of computing experiences. ”
Intel's AI PC Acceleration Program will provide software partners with engineering software and resources to enable AI features for more than 100 million PCs by 2025.
PC AI applications ushered in a "hundred flowers blooming".
At the scene, 23 well-known domestic ISV software manufacturers showcased AI applications jointly created with Intel. Microsoft demonstrated Studio Effect's "eye contact", "auto-framing" and other features. Baichuan Intelligent launched 7 large models in eight months. In addition, Intel not only cooperates with Alibaba Cloud Tongyi Qianwen, RWKV, Xinger AI, Da Pie AI and other companies to enlighten creativity through AIGC functions such as Wensheng Wen, Wensheng Diagram, Wensheng**, AI voice changing, etc., but also creates AI cutout, intelligent recognition, AI motion capture, AI gesture recognition and other functions with applications such as Jianying, Xuanmi, and iQiyi, bringing a more convenient and simpler AI experience to users' daily life.
In addition, Intel has also cooperated with Peking University to achieve a smooth user experience through the multilingual** large model codeshell developed by the Knowledge Computing Laboratory of Peking University. Other ISV partners include: Zhipu, Youdaozi, and Erinie-35. OPPO AndesGPT, Wuwen Core Dome, MiniMax, Sobei, Tencent Meeting, WPS AI, New Aotemei Photography, Filmora Filmora, Jida Technology, Kirin, UnionTech.
Gao Yu, general manager of Intel's China Technology Department, believes that in the AI era, AI applications on the PC side will usher in a big explosion like on mobile phones, "Intel must prepare for the underlying computing power, prepare the tool chain, and integrate with the most active AI ecological chain, so that new applications can quickly migrate from the cloud to the end, and Intel will be the accelerator for this to happen." ”
Let some artists get up first
Wang Lei, Dean of the School of Animation and Digital Arts of Communication University of China, believes that AI is becoming an unprecedented powerful extension of the artist's brain, eyes and hands, and AI technology is good at creating a large number of details, making up for the shortcomings of traditional animation that it is difficult to express grand scenes.
The "Identity Mirror" created by Wu Qiong's team at the Academy of Arts and Design, Tsinghua University, interacts through large language models and AI face fusion technology, leading people to think about the identity relationship between people and their digital selves in the era of artificial intelligenceIn the work "Fake Mirage, Iteration 1, Consciousness Deception", which cooperated with Feng Mengbo, Dean of the School of Experimental Art and Technology of the **Academy of Fine Arts, the computing power of AI PC promotes the reasoning of large language models to realize the artistic expression of the workIn cooperation with Wu Hongliang, dean of the Art Academy of Beijing Fine Art Academy, AI will be used to reproduce Qi Baishi's artistic style for the creation of essays and drawings.
Lv Dong, general manager of Intel's marketing department, said: "Intel provides AIGC art creators with a generation solution supported by local computing power, making it possible for 'everyone is an artist', allowing some artists to get AI first, so that artistic productivity can be better liberated, and laying a technical foundation for the prosperity of AIGC art creation." In addition, the team of Wu Qiong from the Communication University of China, the Academy of Arts and Design of Tsinghua University, and the School of Experimental Art and Technology of the ** Academy of Fine Arts also announced the AIGC Joint Practice Center project to be established and launched by Intel and the three art academies to accelerate the cultivation of AIGC artistic talents.
Core Ultra processor: You can also play triple-A masterpieces on thin and light laptops
Intel Core Ultra processors integrate proprietary AI acceleration capabilities into CPU, GPU, and neural network processing unit (NPU) architectures, making them the most high-performance and power-efficient client processors for AI in Intel's history.
The experience of an AI PC is greatly affected by the performance of the software it is equipped with, and it is stronger when it is stronger. As part of the AI PC Acceleration Program, more than 100 independent software vendors (ISVs) and more than 300 AI acceleration features will be optimized for Intel Core Ultra processors, providing a rich set of AI applications and frameworks.
The Intel Core Ultra processor is the first processor built on Intel 4 process technology and represents the biggest evolution of Intel architecture in 40 years. It uses advanced FoverOS 3D packaging technology to optimize the performance of the relevant modules. The new Performance Core architecture improves instruction execution per clock cycle, while the E-Cores and LP E-Cores, which provide scalable multi-threaded performance and offload more background workloads for the P-Cores, enabling more efficient multitasking performance.
The Intel Core Ultra processor also integrates an Intel Arc GPU, has up to 8 XE cores, and supports the AI-based supersampling technology XESS and DX12 Ultimate, delivering up to 2x the graphics performance of its predecessor. The GPU supports modern graphics features, including hardware-accelerated ray tracing, mesh shading, **1 codec, HDMI 21 and DisplayPort 21 20g。
The latest NPUs (Neural Network Processing Units) are designed for AI workloads that run for long periods of time at low power consumption, complementing the AI capabilities of CPUs and GPUs, capable of bringing 25 times more energy efficient.
Fifth-generation Xeon: Strengthen AI capabilities such as large models
When AI has been deeply integrated into e-commerce, ** and other fields, the demand for AI computing power from leading cloud computing vendors such as JD Cloud, Intelligent Cloud, Alibaba Cloud, and Volcano Engine has become more urgent.
This is the second release of Intel's Xeon Scalable processor family in a year, and the 5th Gen Intel Xeon Scalable processors deliver higher computing power and faster memory than the previous generation.
"The 5th Gen Intel Xeon Scalable processors for AI acceleration are designed to deliver higher performance for customers deploying AI in cloud, network, and edge environments," said Sandra Rivera, executive vice president and general manager, Data Center & AI Group, Intel. ”
"Intel is committed to enabling more customers to design, deliver, and deploy innovative solutions for cloud computing, networking, edge, and massive access services through a comprehensive product portfolio, optimized software, diverse tools, and extensive ecosystem," said Baoli Chen, vice president and general manager of Intel Data Center and AI Group in China. ”
5th Gen Intel Xeon Scalable processors are equipped with AI acceleration on every core to handle demanding end-to-end AI workloads without the need for discrete accelerators, including up to 42% faster inference performance for large language models with up to 20 billion parameters and sub-100 millisecond latency.
5th Gen Intel Xeon Scalable processors deliver up to 21% better overall performance for general-purpose computing and up to 36% faster performance per watt across a range of customer workloads than the previous generation. For customers who follow a typical five-year refresh cycle and upgrade from a previous-generation processor, the total cost of ownership is reduced by up to 77%.
The fifth-generation Intel Xeon Scalable processors are pin-compatible with the previous generation, and large original equipment manufacturers (OEMs) including Cisco, Dell, HPE, Inspur Information, Lenovo, Supermicro Computer, etc., will offer a variety of single-core and dual-core processors for the market to choose from in the first quarter of 2024, while mainstream CSPs will gradually roll out instances based on this new product next year.
Intel Trusted Domain Extensions (TDX) provides isolation and confidentiality at the virtual machine (VM) level for enhanced privacy and data management. As new products are introduced, the feature is available to all OEMs and CSP solution providers.