Semiconductor giant Xilinx showcases its smart car layout in China for the first time: accelerating the evolution of car intelligence from the bottom up

Today, foreign media reported that chipmaker AMD has submitted its plan to acquire US semiconductor giant Xilinx to the European Commission for review, and the deadline for the review is tentatively set for June this year.

This acquisition, which has influenced the global chip industry landscape since it was announced last year, is coming to an end.

Of course, it still needs approval from the regulatory authorities in the UK and China.

Once the acquisition is completed, AMD will compete fully with Nvidia and Intel in the fields of artificial intelligence and the Internet of Things.

On May 25, Xilinx, a semiconductor company that has influenced the global chip industry, appeared at the Guangzhou Automotive Technology Exhibition together with its partners Intron Technologies and Snow Lake Technology.

Xilinx's layout in the field of smart cars was demonstrated in depth in China for the first time.

The significance of joining hands with partners is that partners can develop more concrete products through Xilinx's product empowerment.

For ordinary Chinese car consumers, Xilinx is not a well-known brand. However, for the semiconductor industry, Xilinx, a company that rarely has consumer-grade products, has slowly come to the fore since the acquisition news with AMD last year.

Xilinx is the world's leading supplier of complete programmable logic solutions.

Xilinx develops, manufactures and sells a wide range of advanced integrated circuits, software design tools, and IP (Intellectual Property) cores as predefined system-level functions.

Customers use Xilinx and its partners' automated software tools and IP cores to program devices to complete specific logic operations. Xilinx was founded in 1984. Xilinx pioneered the innovative technology of field programmable logic array (FPGA) and launched its first commercial product in 1985. Currently, Xilinx meets more than half of the world's demand for FPGA products.

In short, Xilinx can significantly improve the efficiency of a large number of different data processing, and has sufficient flexibility, scalability and security.

This is the first time that such a semiconductor company has appeared at the automotive technology exhibition. What identity will it use to enter the field of smart cars?

Xilinx's Core Capabilities

With the improvement of intelligent driving capabilities, more and more data from sensor phones need to be processed by vehicles, both two-dimensional and three-dimensional, which requires sufficient computing power to support.

As Moore's Law becomes increasingly difficult to implement, it is becoming increasingly difficult to increase computing power. In addition, due to the limitations of automobile power consumption, how to process data more efficiently has become a major challenge facing the industry.

This is the capability of Xilinx, or the Xilinx adaptive computing platform, which can adaptively switch between CPU fast computing and GPU heavy-load computing to achieve the highest efficiency state, allowing the computing power required for the vehicle's autonomous driving and intelligent networking to continue to be in the most efficient state.

Among the current successful cases, driving innovation, driver assistance systems ADAS, cabin experience and autonomous driving AD are the three areas of innovation that Xilinx focuses on.

In the driver assistance system, including forward vision, rear vision, radar, AI, collision avoidance, and lane perception fusion, these are all areas supported by Xilinx.

In terms of cabin experience, the covered applications include automatic driver status monitoring, whether there are any items left in the entire cabin, and safety testing of the entire cabin.

Currently, Xilinx's products have been adopted by various forms of smart car hardware. Xilinx calls this cooperation empowerment.

像在车规级固态激光雷达等多种形式的ADAS激光雷达产品上，中国知名的无人驾驶科技公司小马智行（Pony.ai）团队对赛灵思技术支持的自动驾驶传感器融合方案表示：借助赛灵思的技术，他们能够实现传感融合输出的零差异同步，以实现道路上的实时反应，这被认为是前所未的的“显著改进”。

This empowerment was also reflected in the two partners that exhibited with Xilinx at the Guangzhou Automotive Technology Expo.

Xilinx has been in the smart car industry chain for a long time

The two companies are Yingheng Technology and Snow Lake Technology.

Yingheng is a company in the traditional automotive industry. It established a partnership with Xilinx three years ago and has now implemented the domain control function on Xilinx's FPGA products. Snow Lake is a newly established technology company that focuses more on emerging fields. They have discovered the value of the Xilinx platform in the field of intelligent driving, so in V2X, there are product applications both on the road side and on the vehicle side.

In an interview with Zhijiajun, the heads of the two companies said that they both relied on the Xilinx platform and launched different products in different directions.

Chen Ming, general manager and executive director of Yingheng Technology and general manager of Jinmai Electronics, believes that his company focuses on developing the basic software and basic underlying things before moving up to the application layer. The entire solution currently being developed can realize valet parking and autonomous vehicle parking. That is, after the person gets off the car, the vehicle will automatically find a parking space to park, which can be fully autonomous. This technology has now been implemented.

Snow Lake Technology focuses more on the lidar algorithm level.

When talking about the route of autonomous driving, Zhang Qiang, founder and CEO of Snow Lake Technology, said that LiDAR must be installed in the car. He said: "The goal of autonomous driving must be to surpass humans. Tesla uses a pure vision solution because humans do not have LiDAR, but people have very good feedback, so he believes that the end point of AI is to be able to do it like humans. This is the end point of AI. That is to say, since humans do not have LiDAR, they can drive the car very well with pure vision, so they must be able to drive the car very well. But we believe that the goal of autonomous driving must not be to reach humans, but to surpass humans."

Three questions for Xilinx: Advantages, AMD, and Security

On May 25, Zhijiajun had a conversation with Tang Xiaolei, vice president of core market sales for Greater China at Xilinx, on three core issues, namely, the differentiated core advantages of Xilinx, the latest status of AMD's acquisition, and how to protect the underlying security of data.

Tang Xiaolei, Vice President of Core Marketing Sales, Xilinx Greater China

Zhijiajun: Compared with the SOA smart car architecture launched by traditional automotive suppliers such as Aptiv, what unique advantages does Xilinx have?

Tang Xiaolei: Our definition is completely different from that of Aptiv. After all, Aptiv is positioned as Tier 1, and it wants to provide a real solution to OEMs (original equipment manufacturers). Xilinx has always been positioned as a supplier. We have such an architecture, but how to actually implement it? We need partners to make a basic product to provide to Tier 1 or OEM. Relying on the business model of these OEMs, some are willing to do it all by themselves, and some traditional car manufacturers need Tier 1 and Tier 2 to cooperate, but we are a semiconductor supplier, so our architecture is based on our understanding of this market. However, whether it can really be made into an automotive-grade product module and mass-produced on the vehicle still requires the help and cooperation of ecological partners, so our market positioning is completely different.

Zhijiajun: AMD is now competing with Nvidia and Intel. Last year, AMD announced the acquisition of Xilinx. AMD has a strong advantage in the APU field. After Xilinx was acquired by AMD, how will AMD help improve Xilinx's hardware?

Tang Xiaolei: We have always believed that the world will be a world of heterogeneous computing in the future. AMD itself has a CPU and a GPU. If the acquisition is successful in the future, we will add FPGA. In this way, we will basically gather several major elements of heterogeneous computing. We believe that when everyone is together, there will be a synergy effect, that is, a synergy effect, an effect where 1+1 is greater than 2.

Zhijiajun: Not long ago, 360 and Nezha announced a partnership to promote the network security of smart cars. Earlier, NVIDIA launched the RTX3060 product to limit mining. It limits mining computing power through dual encryption of hardware and software, but it was cracked by hackers just a few days after its launch. You just said that Seres' hardware encryption is the strongest. How should this sentence be understood?

Tang Xiaolei: Is there absolute security in the world? Of course not. Including the hardware encryption proposed by NVIDIA that you just mentioned, we do not think that NVIDIA has really made any changes to the hardware. They just added some things to its tools or added some things to the chip. This is different from the chip encryption and hardware encryption we are talking about. We really have a module that can make this hardware to the extreme from the perspective of DNA. We are more confident in our own hardware encryption, which is different from those defensive encryptions.

Xilinx's core positioning is not an automotive supplier, but due to the current situation, it has already penetrated into the grassroots level of smart car development.

Xilinx has unknowingly become a Tier 2 supplier in the automotive industry.

However, in the process of automobile digitization and intelligence, Xilinx has quietly become the foundation of the nerves and brain of the car, which is of course the foundation of all future intelligent functions.

Xilinx can achieve this, which is inseparable from the four capabilities of Xilinx: high efficiency, high flexibility, high scalability, and stability and security.

With Celeron entering the AMD camp, the bottom-level competition in the era of smart cars has undoubtedly begun to take shape among three major players: AMD, Nvidia and Intel.