In the past two years, the sudden emergence of ChatGPT has led to a surge in the popularity of artificial intelligence (AI), which has rapidly penetrated the application end and driven the prosperity of companies such as NVIDIA, AMD, and Intel, and even Broadcom in the data center chip field. Specifically, in terms of terminal side, the smartphone market with the largest user base has become the first foothold for AI, and a mobile phone product form called AI Phone has attracted chip manufacturers such as Qualcomm and MediaTek to exert their efforts.
In addition, another important terminal product, the PC, is also accelerating towards AI. Especially after Microsoft released the "Copilot+ PC" in the first half of this year, AI PCs quickly became popular. According to Canalys, by 2025, the shipment of PCs supporting AI will exceed 100 million units, accounting for 40% of the total PC shipments. The compound annual growth rate in the next few years will be as high as 44%.
Advertisement
Because of such a huge market opportunity, traditional PC chip giants Intel and AMD are accelerating their entry. However, unlike the chip market for AI phones, which has already been determined, a new chip battle has already begun in the AI PC market.
New changes in PC chip, this chip technology rises with the trend
As everyone knows, over the past few decades, the PC chip market has been mainly dominated by two X86 chip suppliers, Intel and AMD. On the one hand, this is because they have been deeply cultivated for decades and have accumulated a lot in chip design. More importantly, over the past few decades, through cooperation with Microsoft, the well-known Wintel (Window+Intel) alliance has created an alliance that seems to be as strong as a fortress.
However, with the launch of Apple M1 on November 10, 2020, and the full transition of its Mac computers to self-developed M series Arm chips, the situation of PC chips has been broken. Arm CEO Rene Haas even said earlier that he aims to occupy more than 50% of the Windows PC market within five years. In view of this, the traditional computing chip giant Qualcomm has strengthened its strength by acquiring Nuvia and exerting its strength in the Arm PC chip market, and MTK is also eager to try. With the arrival of AI, the PC chip market is accelerating the change.
During this period, a local chip company called this chip technology has quietly risen.
Mr. Sun Wenjian, the founder and CEO of this chip technology, recently recalled at the launch of the company's first AI PC chip "this chip P1": "In order to launch this chip, the company's team has been working hard for three years," "Three years ago, when I saw the vigorous development of artificial intelligence technology and saw the market's desire for high-performance CPUs, I had the idea of establishing an AI CPU company." Sun Wenjian continued to say.
Sun Wenjian, founder and CEO of this chip technologyAs a veteran with over 20 years of experience in the design of large-scale digital integrated circuits, Sun Wenjian has served as the Director of AMD chip design and the head of the customer customization department in China, leading an end-to-end R&D team to deeply engage with key global customers, and completing the definition, design, and mass production delivery of several industry-leading smart CPU chips. It is precisely because of such rich experience that he has the confidence to establish Core Technology.
He further pointed out that there is another important reason for choosing such a track. After decades of development, the talent development and industry development of the domestic semiconductor industry have become very rich, and many semiconductor companies are also excellent, capturing a lot of the chip market. However, in terms of CPUs, due to the high barriers of talent, capital, and ecology, it is still in the process of tackling.
But with the industry shifting from traditional CPUs to AI CPUs, Sun Wenjian saw an opportunity for change. More importantly, as Sun Wenjian said, people do not like closed ecosystems. This has brought new opportunities to the industry.
Based on the above insights, coupled with the observation that there is a huge PC market in China, with a complete AI PC industry chain; in the fields of complete machine manufacturing, software design, large model development and application, China is basically at the international leading level; even in some areas with gaps, through the joint efforts of the industry chain upstream and downstream, it will further narrow the gap and other natural advantages, Core Technology was officially established in 2021. And has received financial support from Lenovo Ventures and NIO Capital.
As a company that gathers industry veterans in chip design, software ecology, and terminal applications at home and abroad, Core Technology uses innovative and forward-looking ideas to design and develop advanced high-energy efficiency, intelligent general-purpose CPUs in the industry, empowering the fields of personal computing, vehicle applications, and the construction of the metaverse infrastructure. Core Technology is committed to providing society with low-power intelligent computing power solutions, and is working with leading global commercial customers and ecological partners to build an open ecology based on a unified computing power foundation, creating a new paradigm of intelligent chip 2.0.
With the release of "Core P1", this local AI PC chip company officially entered a new stage.
The release of 6nm heterogeneous chips, the new generation of AI PC computing power base
It is reported that the first chip released by Core Technology, "Core P1", is built with advanced 6nm process technology, integrating a series of leading processors including CPU, GPU, DPU, and NPU, providing rich AI heterogeneous computing resources.
In the core CPU part, "Core P1" is designed with Arm big.LITTLE technology, featuring a 12-core design with 8 performance cores and 4 energy efficiency cores, with the highest main frequency reaching 3.2GHz and multi-level cache design optimized for PC scenarios; at the same time, it integrates 2 SVE2 vector acceleration units, which can achieve enhanced machine learning instructions.
In terms of GPU, "Core P1" integrates a 10-core GPU processor, which can meet the needs of ultimate desktop rendering and general AI computing. It is worth mentioning that this GPU is also equipped with a new generation of hardware ray tracing, which can provide a gaming experience comparable to the host level; its new geometric graphics processing process (delayed vertex shading DVS) can save more than 40% of power consumption, and flexible variable rate shading (VRS) can achieve more than 50% performance improvement. With the desktop GPU software stack for multiple scenarios, this chip can also meet more needs of industry applications.With the built-in Neural Processing Unit (NPU), the "This Core P1" has created a powerful heterogeneous AI engine.
Sun Wenjian also summarized that, in terms of computing power, the CPU of "This Core P1" is equipped with the second-generation Scalable Vector Extension Unit (SVE2), which can support new AI instructions and enhance high-performance computing capabilities; the built-in GPU supports OpenCL and Vulkan acceleration; the NPU computing power can reach up to 30 TOPS, and its dedicated computing power for video and audio processing is faster and more efficient.
Thanks to this leading heterogeneous configuration, "This Core P1" can provide 45 TOPS of edge AI heterogeneous computing power, support the deployment of large models with up to 10 billion parameters on the edge, and run Large Language Models (LLMs) at a rate of more than 30 tokens per second, providing edge AI support for multiple scenarios such as computer vision, natural language processing, and generative AI. With the support of such a rich AI computing engine, coupled with the NeuralOne AI software stack of This Core, developers can be empowered with convenient and efficient heterogeneous scheduling capabilities. This allows "This Core P1" to easily perform AI computations in various scenarios, including computer vision, natural language processing, and generative AI. It is worth mentioning that the generative AI large model of this chip can support models with up to 10 billion parameters, and the large language model can run at a rate of more than 30 tokens per second, supporting the Stable Diffusion text-to-image effect.
In addition to the aforementioned leading computing capabilities, "This Core P1" also has a comprehensive security engine, a variety of peripheral interfaces, and support for multiple operating systems; at the same time, the chip also has a powerful multimedia engine that can support 4K120 frame display, 8K60 frame video decoding, and 8K30 frame video encoding; moreover, the chip integrates a high-performance memory subsystem with a 128-bit LPDDR5 low-power memory, with a capacity of up to 64GB, a data transfer rate of up to 6400Mbps, and a bandwidth of up to 100GB/s; in addition, "This Core P1" can also provide precise dynamic frequency and voltage adjustment, multiple power domains, dynamic power gating, and standard PC power supply working modes, achieving efficient power consumption management.
From the above performance parameters, we can intuitively see the hard strength of "This Core P1". But in fact, behind this chip, there is also a lot of thought from the This Core Technology team. For example, unlike most chip companies that launch a chip only for one application market, This Core Technology positioned "This Core P1" as a "multi-purpose chip" in the product definition - facing both the global and local markets, building an edge AI ecosystem, taking the lead in focusing on the AI PC field, and creating a new generation of AI PC computing power base.
To achieve this goal while avoiding resource waste, it is necessary for the company's team to make a better balance and trade-off when defining the product. This is also a manifestation of the ability of the This Core Technology team.
"The rich functions of This Core Technology chips greatly meet the needs of customers in multiple scenarios; on the other hand, through landing in multiple scenarios, the sales volume of products increases, which spreads the R&D costs of the product, bringing customers a high cost-performance product experience." Sun Wenjian summarized. He also pointed out that in order to meet the different needs of different applications for chips, the company will also add or reduce the chips to make them more suitable for the needs of the terminal.
This also constitutes the computing power base of This Core Technology.
"Three integrations" strategy to create a prosperous ecosystemCo-founder and Vice President of Software Engineering of Core Technology, Liu Gang, frankly stated at the launch of "Core P1": "Hardware is like our body, while software endows a chip with a soul." This is a golden rule for any chip company. In many computing chip companies, the number of software team members often leads the hardware team, which just verifies Liu Gang's statement.
In the Arm PC field, the importance of software or ecology is particularly significant.
Unlike many To B applications that do not need to face consumers directly, devices such as PCs and smartphones are productivity tools for their users, with high usage frequency and long duration. After years of accumulation, the ecology and applications built on their own chips and systems also make it difficult for cross-border players to find opportunities. For example, in the mobile phone industry, Intel made many attempts in the early years, hoping to promote its X86 chips to the mobile field, but in the end, it was all in vain.
Similarly, the Arm chip's journey to the PC has also been through multiple attacks, failures, and then attacks again. In recent years, Apple, which has a completely closed ecosystem, has finally opened up the situation. This also alerted Microsoft to increase its investment in "Windows on arm" and give the Arm PC ecosystem practitioners a reassuring pill.
With the arrival of AI, new opportunities for AI PCs are surging. In order to make Arm-based chips go to AI PCs, Core Technology co-founder and Vice President of System Engineering, Chu Ranzhou, introduced the "three integrations" (integration, integration, and permeation) strategy to create an AI PC platform solution from a new perspective.
In terms of "integration," it means that when the company defined "Core P1," it started from the beginning to let go of all the baggage, integrate and absorb the strengths of X86 and Arm, and perfectly realize the popularization and application of Core Technology Arm architecture CPU chips in the PC field; "integration" means that the company hopes to quickly integrate into the PC industry chain, make full use of China's advantages in the PC and mobile device fields accumulated over the years, and work together to provide consumers with high-quality AI PC products; "permeation" means that Core Technology will keep up with the development trend of public large models and local large models, and provide excellent end-side platforms that integrate cloud computing infrastructure, truly making AI benefit everyone.
With the support of this "three integrations" strategy, Core Technology has launched the AI PC platform solution of the P1 chip, featuring scalable heterogeneous computing, support for multimodal human-computer interaction, high-bandwidth storage, and platform-level security shields. In response to the current situation of Arm PC chips and AI PC chips, Core Technology focuses on four major directions: boot firmware, kernel, graphics acceleration, and AI solutions for full-stack software innovation.
At the boot firmware level, Core Technology has achieved a significant breakthrough in supporting multiple operating systems with a set of firmware and supporting both ACPI and Device Tree specifications with a set of Linux kernels; in order to enable the Arm GPU to also achieve the ultimate user experience on the PC side, Core Technology has independently designed the Core GO graphics engine, which introduces an application compatibility layer and achieves original optimization at the core driver level, adapting to various mainstream desktop environments, compatible with traditional applications, supporting the OpenGL standard, and coordinating with different multimedia frameworks, solving the industry's pain points in one stop.
Faced with many problems and challenges in the deployment of generative AI on the end side, Core Technology will launch the NeuralOne AI software stack in the future, providing support for heterogeneous AI accelerators to meet the diversity and complexity of end-side inference requirements; at the same time, it provides a unified NeuralOne API to hide specific hardware details and reduce the difficulty of application programming. In Core Technology's view, for the fragmentation of models and inference frameworks, Core NeuralOne provides a unified SDK to meet the needs of multiple engines and support a wide range of model formats.
In order to better promote the construction of the AI PC industry chain, during the launch of "Core P1," a special AI PC industry chain strategic cooperation launch ceremony was also held. Core Technology, together with Lenovo Group, Arm Technology, Tongfang Dingxin, Wanmos, Tongxin Software, Kylin Software, Jiangbolong, Bai'ao Software, Wuwen Xinqiong, and many other partners, will give full play to their respective advantages in IP design, full-stack software development, and system design, to accelerate the commercialization of domestic AI PC innovative products and promote the industry to achieve sustainable development.As mentioned at the beginning of the article, in the new track of AI PCs, not only are there traditional giants, but newcomers like Qualcomm in the Arm PC field are also rising rapidly. For example, in the "Copilot+ PC" released by Microsoft in the first half of the year, Qualcomm's chips were unprecedentedly chosen, which is enough to witness the strength of this giant.
In the interview, Sun Wenjian told the author that the development of AI technology has brought about the opportunity for change and entrepreneurship, and the birth of the AI PC track in China is timely, becoming a new force on the domestic AI PC track. It is worth mentioning that X-Core Technology is also the first AI PC chip supplier in China and one of the few core partners of Arm in the global PC ecosystem.
Sun Wenjian frankly stated that it is very difficult for a single company to create a huge Arm PC ecosystem. "The entire ecosystem needs to be connected from IP to chips, to complete machines, to operating systems, to large models, to applications. Under this situation, everyone has already reached a consensus and actively taken action. I think everyone should form a complementary ecosystem, exert their advantages in the construction of the new ecosystem, and truly optimize each link of the Arm-based AI PC, hand it over to customers, and form a huge industrial chain win-win."
From this, it can be seen that in the era of AI PCs, Chinese chip manufacturers are bound to play an important role.
Comments