congratulations! Geely Automobile self-developed 7nm car gauge chip


Release time:

2021-11-10

Recently, Geely announced that the "smart cockpit chip SE1000" self-developed by "Sinengine" adopts a car-grade 7-nanometer process, multi-core heterogeneous, advanced manufacturing process, and is China's first 7-nanometer car-grade SoC chip.

congratulations! Geely Automobile self-developed 7nm car gauge chip

Recently, Geely announced that the "smart cockpit chip SE1000" self-developed by "Sinengine" adopts a car-grade 7-nanometer process, multi-core heterogeneous, advanced manufacturing process, and is China's first 7-nanometer car-grade SoC chip.

Recently, Geely announced that the "smart cockpit chip SE1000" self-developed by "Sinengine" adopts a car-grade 7-nanometer process, multi-core heterogeneous, advanced manufacturing process, and is China's first car-grade SOC chip with a 7-nanometer process. This chip will be mass-produced next year after the vehicle-level certification is completed.

From 2024 to 2025, Geely Automobile will successively launch 5-nanometer vehicle-mounted integrated supercomputing platform chips, as well as high-computing autonomous driving chips, with a computing power of 256TOPS. By 2025, realize the commercialization of L4 autonomous driving and fully master L5 autonomous driving.

The following is the content of the speech on chips in the "Smart Geely 2025" strategy released by Geely Automobile Group CEO Gan Jiayue:

Just a few days ago, a small chip "Smart Cockpit Chip SE1000" was delivered to my office.

Although the "chip" is small, only he can support the ecological network of smart technology.

 

 

This "chip" is self-developed by "Xinengine" technology, which adopts a car-grade 7-nanometer process, multi-core heterogeneous, and advanced manufacturing processes.

This chip is only 83 square millimeters, but it integrates 87 layers of circuits, 8.8 billion transistors, and lights up at one time. To use an inappropriate analogy, it is equivalent to building 8.8 billion rooms in a place as large as a fingernail, and 87 floors are built on it. This is China's first 7-nanometer automotive-grade SOC chip.

This chip will be mass-produced next year after the vehicle-level certification is completed.

 

 

After the 7-nanometer SOC chip, from 2024 to 2025, we will successively launch 5-nanometer vehicle-mounted integrated supercomputing platform chips, as well as high-computing autonomous driving chips, with a computing power of 256TOPS.

Meet the needs of L3 intelligent driving. At the same time, through the multi-core combination, the computing power can be expanded to meet the computing power requirements of higher-level autonomous driving!

Automobile factory announced self-developed chips 

The shortage of chips is intensifying on a global scale, which continues to constrain the production capacity of auto manufacturers. More and more auto companies have announced self-developed chips.

Hyundai Motor

In October, Jose Munoz, Hyundai's global chief operating officer, announced plans to develop its own semiconductor chips to reduce its dependence on chip makers.

While Hyundai's sales did not suffer much in the last quarter, Munoz did say the "toughest months" were August and September, Reuters reported. The South Korean automaker had to temporarily close some plants, but Munoz said the worst of the chip shortage was over, citing heavy investment by Intel to expand capacity.

Still, Munoz told reporters that Hyundai does not want to be left without supply again and that the company needs room to be more self-reliant. He acknowledged that developing the chips in-house would take a lot of time and investment, but that it was "something we're working on." The chips are most likely to be made at Hyundai Mobis, the former's components subsidiary.

Saic GM-Wuling

In September this year, SAic-GM-Wuling publicly stated that it was accelerating the "strong core" strategy. Wuling chips were unveiled to the public, and the company aimed to achieve over 90% localization rate of chips for GSEV (global small pure electric vehicle architecture) platform models during the 14th Five-Year Plan period.

Saic-gm-wuling also said that in the fields of 5G communication chips, memory chips, energy chips and artificial intelligence chips, it will work with outstanding domestic partners to jointly break through core technologies and accelerate the pace of chip localization.

The domestic chip partner of Wuling chip is Xinwang Micro, which uses the ChipON developed KungFu32 kernel architecture to process the KF32A15X series. According to Xinwang, the processor uses a three-stage pipeline, 16 bit /32 bit mixed instruction set, the KF32A15X main frequency is 120Mhz, Flash up to 512KB. ChipON has also developed its own development tools, including an integrated development environment, a C compiler and an emulator. Truly realize the full autonomy from the chip to the tool chain.

Tesla

In August of this year, Tesla released a custom chip for fast training artificial intelligence models in data centers. Ganesh Venkataramanan, senior director of Tesla's assisted driving AutoPilot hardware, introduced the AI ​​chip called D1, which uses a 7-nanometer manufacturing process and has a processing capacity of 102.4 billion times. A set of chips can provide a computing power of 9 petaflops.

"We are assembling the first chipset." He said, "Tesla's chip will become the fastest computer for AI model training." Video data captured by cameras inside Tesla vehicles.

Tesla CEO Musk said that the company's Dojo project will start production next year. As early as a few years ago, Tesla stated that it would abandon Nvidia's AI autopilot chip and asked the team to develop an ultra-fast chip for computer training, which is also the origin of the Dojo project.