In 2021, Tesla Model S Plaid is equipped with AMD's Ryzen RDNA 2 processor, and then gradually replaces the Intel A3950 processor for the Model 3 and Model Y high-end versions of the in-vehicle infotainment system. After completing its acquisition of FPGA company Xilinx in February, does AMD, an American semiconductor company, have a clear automotive strategy like Qualcomm and Nvidia? Will there be a 5-10 year plan to really become an important player in the automotive chip market? This thing is really interesting, in the era of high computing power, cross-field playing often comes very suddenly.
Figure 1 AMD's Ryzen™ embedded APUs
Part 1: FPGAs in cars
The demand for high-performance SoCs in smart cars is growing steadily, and AMD's acquisition of Xilinx is aggressively expanding into the automotive space (which has become AMD's FPGA division), focusing on FPGAs, adaptive SoCs, and software roadmap work. Traditionally, FPGAs have been temporary applications in automotive systems, but as iterations have accelerated, Xilinx's FPGAs have been designed into ADAS embedded controllers to process data from cameras, 4D imaging radar, and lidar.
Figure 2 The FPGA has a bit of a "hand" in ADAS
From the perspective of specific applications, Xilinx has partnered with Seeing Machines to provide a semi-customized version of its car-grade chip for Seeing Machines' Fovio chip. Subaru used Xilinx's FPGA scheme. Xilinx is also allied with Veoneer, where xilinx FPGA-based, previous-generation Bosch and Nidec's front-view cameras are still in production.
Figure 3 Multiprocessor SoC FPGAs used in the Subaru Eyesight system
Xilinx, the premier chip supplier for processing surround view camera data, includes products from Bosch and Magna, which are already in production. Several other large Tier 1s will come into production next year.
Due to its high degree of freedom, Xilinx dominates the processors of emerging 4D imaging radars (accounting for 85%-90% of the small-scale innovation market), and the main competitor to this piece is NXP.
Figure 4 XILINX's 4D millimeter-wave radar scheme
In the field of lidar, due to the rapid iteration of a large number of start-up companies, the same customers are based on Xilinx's chips for design and use, Innoviz self-developed this piece, in general, the initial use of FPGAs is more.
Figure 5 Lidar and XILINX
In the world of advanced autonomous assisted driving, Xilinx Automotive (XA) platforms play a key role in powering autonomous driving modules to enable high-speed data aggregation, pre-processing and distribution (DAPD) and computational acceleration. The adaptive XA SoC platform not only optimizes the processing of more and more complex security-critical applications, but also meets the computational latency, performance, power efficiency, and functional safety requirements between sensors and domain controllers. This set of products mainly includes platforms such as XA Zynq-7000 and Zynq UltraScale+ MPSoC.
From a cost point of view, FPGAs are more expensive than other SoCs, but the overall development cost of SoCs is too high, which makes the flexibility of the FPGA platform make up for the higher costs. The current advantage of FPGAs lies in the rapid development of sensor technology. As automotive sensor performance improves, SoC designers are struggling to adapt. When a 1-megapixel camera is upgraded to 2 million, 4 million, or even 8 megapixels, logic devices must keep up with this rapid iteration, and the difference becomes more pronounced as the sensor is further upgraded.
Part 2: AMD-Xilinx Automotive Strategy
From an application perspective, the AMD-Xilinx automotive strategy is a growth area such as domain controllers and local controllers in the automotive system architecture, effectively leveraging AMD's high-performance computing power and Xilinx flexibility. To succeed in the automotive market, chipmakers need not only chips, but solutions that start at the board level.
Figure 6 The architecture of the high computing power era revolves around the chip architecture
Starting with AMD's high-performance processors, the combination of AMD and Xilinx offers top-notch capabilities, including a best-in-class x86, a best-in-class GPU (this entertainment feature is very leading for all publicly available gaming platforms), and a best-in-class adaptive SoC with programmable logic. In the specific automotive requirements, AMD can provide Arm Cortex A or R core to achieve a complete solution, AMD from Xilinx to get the most critical parts of the automotive chip manufacturer, including the relationship with the car manufacturer, the completion of the state of the embedded solution chip and Xilinx software.
Figure 7 ADM enters the automotive market
In this sense, AMD-Xilinx is more like a newcomer to the automotive chip industry, but their synergy is strategically viable, and AMD also recognizes the needs of automotive customers, working on the Arm platform or building Arm cores into their chipsets, and being able to develop some unique solutions.
brief summary:
The automotive era, in a sense, is to differentiate and meet the needs of users as the king, through the acquisition of Xilinx, AMD has a good start in the automotive market. But if AMD wants to continue to fight competitors such as Qualcom, Nvidia and Mobileye in this field, it needs to use its advantages in the field of high computing power to break the development rules of the automotive industry.
Figure | network and related screenshots
About author:Zhu Yulong, senior electric vehicle three-electric system and automotive electronics engineer, author of "Automotive Electronics Hardware Design".
Write a message
2030 Mobility Research Laboratory
The first in China to consist entirely of PhDs
In-depth research organization for new mobility in automobiles
·
Zhihu private message ID: Fish is not fish